Productive Thinking of Gifted Children in Classroom Interaction

DOCUMENT RESUME ED 026 758 EC 002 206 By-Gallagher, James J.; And Others Productive Thinking of Gifted Children in Classroom Interaction. CEC Research MonographSeries B, Number B-5. Council for Exceptional Children, Washington, D.C. Spons Agency-Elizabeth McCormick Foundation, Chicago, Ill.; Office of Education (DHEW), Washington,D.C. Pub Date 67 Note-113p. Available from-The Council for Exceptional Chiidren, NEA, 1201 16th Street, N.W., Washington,D.C. 20036 ($2.00). EDRS Price MF-$0.50 HC Not Available from EDRS. Descriptors-Academic Achievement, *Attitudes, Cognitive Measurement, *Cognitive Processes,Convergent Thinking, Divergent Thinking, Evaluative Thinking, *Exceptional Child Research, FamilyRelationship, *Gifted, HighAchievers,*Interaction,JuniorHighSchoolStudents,ParticipantCharacteristics,Questioning Techniques, Questionnaires, Sex Differences, Student Attitudes, Teacher Attitudes, Tests Identifiers-Guilford, Theoretical Model for Complete Structure of Intel A research project attempted to identify and classify productivethought processes of gifted junior highschool students and their teachers. Subjects were 176 gifted high achieving students of both sexes with a verbal 10 rangeof 127.21 to 136.35, a nonverbal 10 range of 122.88 to 134.59, and a chronological age rangeof 12.50 to 14.63 years. The students were given tests to determine attitudesand divergentthinkingabilitiesand were evaluated forsocialqualities and class contributions by their teachers. Parents were asked tocompleie questionnaires independently to determine family relationships. Three judges, working as a team, tape recorded five consecutive sessions of classes in social studies, science, andEnglish conducted by different teachers inthefall and again the following spring; all comments were classified accordingto levels of thinking definedinGuilford's structure of theintellect.The types of questions asked by teachers strongly influenced the quality of pupil response. More than 507 of questions asked in a class session were cognitive memory questions.The second most frequent category was convergent thinking, with a much smaller proportionof divergent and evaluative thinking questions. (BB)

, ,

Prodifutcetd civehiiThdirnekning -II , 073 c 0.CEC in Classroom mi Interaction

mz 1 c) _ CD W

PROCESS WITH MICROFICHEAND PUBLISHER'S PRICES. MICRO- FICHE REPRODUCTION ONLY. U.S. DEPARTMENT OF HEALTH, EDUCATION & WELFARE PROCESS WITH MICROFICHE OFFICE OF EDUCATION AND PUBLISHER'S PRICES. MICRO- FICHE REPRODUCTION ONLY.

THIS DOCUMENT HAS BEEN REPRODUCED EXACTLY AS RECEIVED FROM THE PERSON OR ORGANIZATION ORIGINATING IT.POINTS OF VIEW OR OPINIONS

STATED DO NOT NECESSARILY REPRESENT OFFICIAL OFFICE OF EDUCATION POSITION OR POLICY.

CEC Research Monograph Series B, No. B-5

Productive Thinking of Gifted Children in Classroom Interaction

James J. Gallagher Associate Director, Institute for Research on Exceptional Children, University of Illinois

Mary Jane Aschner Boston University

William Jenné Oregon State University

The research reported herein was performed pursuant to a contract with the US Department of Health, Education, and Welfare, Office of Education. During the first year, it received support from a grant from the Elizabeth McCormick Foundation. "1"

Research Monographs of TheCouncil for Exceptional Children, aDepart- ment of the NationalEducation Association, are issuedperiodically. Price per single copy: $2; Discount rates:2-9 copies, 10%; 10 or more copies, 20%. Address all inquiries and orders to: TheCouncil for Exceptional Children. NEA, 1201 Sixteenth Street, N.W.,Washington, D.C. 20036. Copyright 1967 by The Council forExceptional Children, NEA Library of Congress Card Catalog No.67-22603

Permission to reproducethis copyrighted workhas been granted to the EducationalResources Information Center (ERIC) and to the organizationoperating under contract with the Office toEducation to reproducedocuments included in the ERIC systemby means of microficheonly, but this right is notconferred to any users ofthe microfiche received from theERIC DocumentReproduction Service.Further reproductionof any part requires permission of the copyright owner. Acknowledgments

In a project as long and extensive asthis one, it is only natural that the authors have many debts ofgratitude to the large number of persons who wereresponsible for its completion. During the course of the project, numerousresearch associates and assistants at the Institute for Research onExceptional Children, Uni- versity of Illinois, served in a multitudeof capacities, such as test scoring, judges for the classification system,and aiding in the anal7sis of the data. These were Sibel Afsar,Helen Farr, Joyce Perry, Mary Ann Rainer, Faye Shaffer, Robert Smith, andElizabeth Thomas. Dr. Betty Teska provided needed assistance on thedifficult scoring problems on the sentence completion test inaddition to her other duties. The author would like to expresshis appreciation to those school administrators who accepted the inconveniencesthat this research im- posed on their school programs. Thesewould be Dr. David Jackson, who at that time was principal ofUniversity High School; Dr. W. L. Shoemaker, Director of the Guidance Department atUniversity High School; Dr. William Cherry, Superintendentof Schools in Li bana; and Mr. Wendell Anderson, Principal ofUrbana Junior High School. The cooperation and enthusiastic supportof these administrators were responsible for the warmreception the research people foundin the schools in which the taperecordings were obtained. A large debt is also due to theteachers who graciously allowed us to intrude on their classrooms so that wecould learn something more about teacher pupil interaction.These teachers were Miss Frances Cottrell, Mrs. Viola Gribinovsky, Mr.Keith Hanson, Mr. Thomas Mor- gan, and Mrs. CarolePalmer. In a real sense, this was a mostimportant first step in removing the cloakof mystery over the classroom actions in the hope that the informationgained would be of help to them and to other teachers. The services of the Digital ComputerLaboratory at the University of Illinois made it possible to do someof the more involved statistical analyses, and Mrs. Sondra Phillipsprovided invaluable assistance in preparing the data for analyses.

"'t2isqt,:.,1(1 .71CIWF,51,7, ,"173.F"9,V.1"75777&771 T.T.E11771M77,747,?,

The author also expresses thanks to numerous fellow members of the staff at the Institute for Research on Exceptional Children at the University of Illinois who graciously listened to many progress reports and gave good counsel and advice along the way. Finally, to those students at University High School and Urbana Junior High School who participated in die tape recordings and in the testing program that were a part of the total project, we wish to give our sincere thanks and appreciation.

JJG Contents

PAGE CHAPTER

Acknowledgments

List of Tables and Figures vii

1 1/ Overview of the Study Objectives and Procedure 1 Results 3 Conclusions and Implications 5

2 / Background of Project 7 Research in Classroom Interaction 8 Research in Cognitive Processes 9

3 / Objectives and Methods 13 Subjects 14 Data Collection 15 Classification System 17 Measuring Instruments 28 Analysis of Data 24

4 / Results 37 Proportion of Various Thought Processes Expressed by Teacher and by Students 37 Relationship between Inc lass and Extraclass Factors 53 Relationship between Cognitive Performance of Teacher and Student 60 Consistency of Classroom Expressiveness in Gifted Children over Change of Time, Subject Matter, and Teacher 61 Daily Consistency of Teacher Student Expressiveness within Class Sections 62 Factors Influencing Type of Production in Class 66 Sex as Relevant Variablein Classroom Expressiveness and Other Comparisons 70 Subgroup Differences inCognitive Styles 75 Differences between Expressiveand Nonexpressive 80 Students Family Data 84

5 / Discussion andImplications 86 Cognitive Structure and Process 86 Project Limitations 95 Future Research Plans 96 A Final Word 99

References 100 KY,

List of Tables and Figures

=MENP.

PAGE TABLES 1. Differences between Laboratory Exercisesand Reality in Use of Productive Thinking 10 2. Chronological Age and AbilityLevel of Students by Group and Sex 14 3.Classes Tape Recorded for FiveConsecutive Sessions for Analysis of Productive Thinking 16 ,f 4. Secondary Classification Categories 22 5. Percentage of Agreement by TwoConsensus Teams Classifying One Session of IDEA 3 27 6. The Stability of Fluency as a PersonalCharacteristic 30 7. Interscorer Valence Agreement forSentence Completion Test 33 8. Teacher Rating Scale on Personalityand Attitudinal Characteristics of Students 35 9. Secondary Category Classificationsof Teacher Questions and Statements 46 10. Secondary Category Classificationof Student Statements 50 11. Group A Principal AxisComponent Loadings by Sex (Varimax Rotation) 54 12. Group B Principal Axis ComponentLoadings by Sex (Varimax Rotation) 55 13. Spearman Rank Order Correlationsof Student Performance 60 versus TeacherPerformance 61 14. Consistency of ClassroomExpressiveness in Gifted Students 15.Differences by Chi Square onProportions of Teacher Cognitive Expression within Five Class Sessions 63 16. Differences by Chi Square onProportions of Student Cognitive Expression within Five Class Sessions 65 17. Analyses of Variance of ThoughtProcesses Expressed by Group, Class, and Day(Teachers) 67 18. Analyses of Variance ofThought Processes Expressedby Group, Class, and Day (Students) 69 71 19. Verbal Classroom Expressivenessin Gifted Groups by Sex 20. Comparison of Boys' andGirls' Adjusted Scores on 72 Classroom Performance 73 21. Sex Differences on TestMeasures and TeacherRatings 22. Cognitive Style Related to Classroom Performance by Adjusted Classroom Scores 77 23. Differences between Cognitive Style Groups by Teacher Ratings on Cognition and Sociability 78 24. Comparison of Cognitive Style Subgroupson Semantic Differential 79 25. Comparison of Expressive and Nonexpressive Gifted Studentson Cognitive Factors and Teacher Ratings by Sex 81 26. Comparison of Expressive and Nonexpressive Gifted Studentson Semantic Differential Scale 82 27. Comparison of Expressive and Nonexpressive Gifted Studentson the Sentence Completion Test 83

FIGURES

1. Theoretical Model for the Complete Structure of Intellect 18 2. Sample Page of Classification Flowchart (from HAT 1 Session) 25 3. Proportion of Thought Processes Asked for by the Teacher (Social Studies) 38

4. Proportion of Thought Processes Given by Boys (Social Studies)... 39 5. Proportion of Thought Processes Asked for by the Teacher (Science) 41 6. Proportion of Thought Processes Given by Boys (Science) 42 7. Proportion of Thought Processes Asked for by the Teacher (Social Studies and English) 43 8. Proportion of Thought Processes Given by Boys (Social Studies and English) 44 1 Overview of the Study

The importance of teacher student interaction in the classroom has long been recognized, but until recently assessment of this area has been entirely the province of supervisors, ill equipped for scientific observation. The only data available consisted of unreliable personal anecdotes. Re- search in the social sciences has too often avoided the complexities of social setting for more precise laboratory studies. Two important developments have changed this situation: advance in technology and changes in the theoretical approach to mental function- ing. The use of tape recorders for making a permanent record of classroom intexaction has provided the technical means for intensive study. The theoretical contributions of Piaget, Guilford, and others, in focusing attention on the cognitive processes, have had extensive impact on both theory and practice in the social sciences.

Objectives end Procedure By identifying and classifying productive thought processes expressed by intellectually gifted children and their teachers, and examining relationships between these thought processes and other variables, the project described here sought specifically to determine:

1. Whether gifted children reveal distinctive individual patterns of cognitive performance. 2. Whether there is a significant relationship between the teacher's cognitive performance, and variations and patterns in the students' thought processes. 3. Whether gifted children who show a high proportion of expressive thought also obtain high scores on tests purported to measure productive thinking. 4. How attitudes and self concepts are related to the various thought patterns expressed in the classroom. 5. Whether significant differences appear among various subgroups of gifted children.

1 6. What relationships exist betweenaspects of family environment and verbal expression in the classroom.

The subjects in this study, 176 high achieving, academically talented students enrolled at the junior and senior high school level in either a university laboratory school or public school, were observed in 12 separate class sections covering social studies, science, and English. The ranges of subgroup means on age and ability were:

CA: 12.50-14.63 Verbal IQ: 127.21-134.35 Nonverbal IQ: 122.88-134.59.

There were no significant differencesamong groups with the exception of one group significantly loweron nonverbal IQ. Throughout the study data on girls and boys were compiled and analyzed separately. Five consecutive one hour class sessions weretape recorded in each of the 12 classrooms, and two observers in each classroom took additional notes. No attempt was made to modify the routine classroom situation. A final typed tapescript reproduced all relevant activity, both verbal and nonverbal, as thoroughly as possible. The students were given a battery oftests measuring both cognitive processes and personality and attitude variables. The Uses and the Con- sequences tests from the Guilford battery were used to measure aspects of intellectual development suchas fluency and intellectual flexibility. Students were given either a modification of the Rhode-HildrethSen- tence Completion Test (Rhode, 1957) or a semantic differential scale previously used by Pierce and Bowman (1960) witha similar gifted population. In addition, results from standardmeasures of achievement and group intelligence tests, as wellas teacher ratings, were used in the final analysis. A family study questionnairewas mailed separately to the mother and father of each child witha request that it be completed independently of spouse. The questionnaire included dataon demographic variables such as occupations, social class, and family size.Instruments used to evaluate this information included Farber's (1962) index ofmari- tal integration; a modified version of Torgoff's (1960)Developmental Timetable; and the Life Goals and Behavior Choices instruments.In order to evaluate parent-child communication, each childwas asked to assess his parents' satisfaction with his behavior on a scale which paral- leled some of the scales in the parents' questionnaires. The Aschner-Gallagher Classification System (Aschner, Gallagher, Perry, Afsar, Jenne, & Faar, 1965), basedon the Guilford model of the

2 structure of intellect, wasthe instrument used to assessthe different classroom. One focus of types of cognitivebehavior expressed in the project staff activities wasthe further refinementand development of this system itself. rotation was used to in- A principal axis factoranalysis with varimax variance was vestigate interrelationships amongvariables. An analysis of included to identify thosefactors accounting forthe greatest variance in the classroom. Standard ttests were used to comparethe subgroups with- computed to indicate pat- in the total population,and chi squares were tern deviationsfrom chance expectationin classroom operation.

Results of the questions Cognitive memory questionsmade up 50 percent or more The second most frequent asked by teachers inpractically all sessions. category was convergentthinking, with a muchsmaller proportion of sessions, requests divergent and evaluativethinking questions. In some evaluative thinking wereentirely absent, and their for divergent and It would presence seemed todepend upon teacherstyle and subject area. almost all classroomdiscussions, regardless ofindividual appear that cognitive memory and teacher style, have asubstantial proportion of convergent thinking as necessarycomponents. follow a different Thought expression inteacher questions seemed to appeared to represent pattern from that inteacher statements.Questions of advancing classdiscussion and wereused to the teacher's method state- elicit the desired typeof response fromthe students, whereas the teacher's individualcognitive style. The statements represented while ments remainedrelatively constant in stylefor each individual, introduced, developed, the types of questionsvaried as the subject was differences among thedif- and concluded inclass. However, significant ferent teachers were notedin both these areas. observed to varysignificant- Each teacher's patternof performance was instances, from one classsection ly from one day toanother and, in some concepts. It wouldbe difficult to another, evenwhen teaching identical teacher's performancewithout indicating: (a)the par- to characterize a working, (b) the ticular group of studentswith which the teacher was teacher's goals for this group,and (c) the degreeof class progress toward these goals at thatparticular time. of teacher questions A close relationship wasnoted between the type asked and the patternof thought expressionobserved in student re- expression were sponses. It wasclear that characterand style of verbal mainly directed by theteacher. of secondary Teacher differences wereobtained over a wide range 3

., , verbal expression categories, a,well as on such noncognitivevariables A close as ratio ofpositive to negative verdictsand expressions of humor. correlation was found betweenteacher and studentperformance in these areas also. significant- The degree of expressivenessof the individual students was ly consistent despite changesin subject matter, teacher,and time. Con- trary to preliminaryexpectations, individualstudents did not ap- who pear to specializein any particular majorthought category. Those performed well in one categorytended to be superior in all.The fact that effective performance in theclassroom required use ofall the major categories seemed to be responsiblefor this high intercorrelation. expressiveness and Sex differences weri,obtained in degree of classroom general attitudes toward self andothers. Boys tended to show more expressiveness in all measurabledimensions in the classroomand more confidence in their own abilities;girls appeared to be morepositive in attitudes toward others andexpressed a more positiveattitude toward the world around them. Sincewritten tasks reflecting cognitiveability revealed no sex differences, orin some instances superiorityby the girls, this verbal superioritywithin the classroom may havebeen related more to personality andattitudinal dimensions than tocognitive ability. IQ score did not seem to be asignificant variable. It wcs notrelated to either classroomexpkessiveness or dimensions ofattitude and personality, but it must be rememberedthat the IQ range in this groupof gifted students was extremelylimited. Performance on divergent thinking tests wasnot related significantly to classroom expressiveness,but, as expected, did seem tobe related to measures of self conceptsand attitudes. In boys, goodperformance on divergent thinking tests seemedrelated to a degree of socialindepen- dence and nonconformity; ingirls, it seemed more related to atotal pattern of good academicperformance and personaladjustment. An attempt was made to replicateprevious work by Getzels andJack- son (1962) andTorrance (1959). Significantdifferences in cognitive styles were found among thehigh/low IQhigh/low divergentsub- groups: (a) Teachersrated boys classified lowIQhigh divergent less well on cognitive activities than highIQlow divergent boys. (b) The high IQlow divergent boys seemed tohave a much less positiveattitude toward the concept of work. Thisraised the possibility that their acceptance of the goal of academicachievement and their drive toward academic attainment (work?) was notwithout its cost in terms of under- lying attitudes. (c) In general, thegirls rated high IQhigh divergent received more favorable teacherratings and performed more expressive- ly in the classroom than any othersubgroup of girls. This supports a general conclusion that girls who aresuperior in IQ and divergent

4 thinking tasks are more self confident and moreexpressive in their academic performance. Incomparing students who were superior in classroomexpressiveness with those inferior in expression, differencescentered more in the attitudinal dimensions than in the cognitiverealm.

Conclusions and Implications Previous observations were essentially confirmedregarding the crucial role played by the teacher as the initiatorand determiner of the kinds of thought processes expressed in the classroom.The individual differences obtained in this limitedsample of teachers, all of whom had been rated as superior instructors, suggest thatteaching styles need to be studied in greater depth to determine the impactof such styles on student performance. While the present classification system ofclassroom interaction appears to be a useful first stepin describing teacher and student behavior and categorizing differences, development oflarger units of measurement is needed to aid in interpreting classroom strategyand interchange. During the study, serious questions were raised about theadequacy of the Guilford model of the structure of the intellect,particularly as it related to intellectual operations. The investigatorsconcluded that divergent thinking and convergent thinkingcall for similar mental operations since both demand the use of logical syllogisms. Thedistinctive characteristic in divergent thinking is the large numberof syllogisms required. Differences obtained between divergent and nondivergentstudents seem to rest in the student'sattitude toward intellectual risk taking, which, in turn, is influenced by personality andattitudinal dimensions. Differences in cognitive style were obtained among varioussubgroups. This suggests that the way in which individualscharacteristically process information received from their environment interactswith their attitudes and, to some extent, influences their classroomperformance. A question was raised in the present study as to whether or notthe individual adept in performing on tests of divergent abilityis, in fact, the creative thinker. As pointed out many times, creativityrequires the mobilization of all kinds of thought operations, each inits proper sequence, to produce the finalcreative product. It appeared that some youngsters who performed well ondivergent thinking tests were merely showing a lack of ability to make a criticalevaluation of their own performances. Such persons not only temporarilydelay evaluation, essential for allowing newly formed ideas a chance todevelop, but postpone it indefinitely. They are not adept at mobilizingtheir scattered intellectual resources to attack a specific problem. Onthese grounds, teachers

5 who rated these students less favorably than the youngsters who showed high IQ but low divergent patterns were not revealing a conformist bias so much as they were accurately reporting the lesser ability of the divergent student to focus on a given assignment. The sex differences which pervaded the present study indicate once again that similar performance on the part of youngsters can conceal quite different patterns of motivation and attitudes. In particular, boys seem to use expressive performance as intellectual aggressiveness, whereas girls use it as part of a total pattern of academic efficiency. However, little is known about the dynamics and motivation of gifted girls, and more analysis and investigation of the personality and motivational structure of gifted girls should be undertaken. Further work is contemplated along the lines of identifying critical incidents which shape and mold the operation of the classroom and the process by which such topics, or larger units of discussion, can be identified and classified in a broader system of cognitive processes. A new model of cognitive processes related to complex social interchange in the classroom is being developed. The results on the family indices and questionnaires, reported in detail by Jenne (1965), were equivocal and showed no clear patterns of parental attitudes influencing child behavior. One interesting specula- tion by Jenne was the possibility that parents wishing to provide a fertile environment for productive thinking in their children would have to place more parental control on boys but less parental control on girls in order to compensate for current cultural biases in child rearing.

6 2 Background ofProject

establish the first time some- It is no doubt ahistorical impossibility to "It's what the teacherdoes that counts."While it would be one said, substantially with such a difficult to find anyonewho would disagree manifestly true that,before the last twodecades, edu- statement, it is detailed study of teacher cational research haddevoted little time to pupil interaction. behavior or to teacher educational field has ma- A generation ofresearch personnel in the that the ideal researchproject is an evaluation tured with the belief paradigm. design carefully patternedafter a biologicalmedical research educational researchprojects, the students are ran- In the most careful The experi- domly distributed intocontrol and experimental groups. of instruction) mental group gets the treatment(often a new program receives nothing or theequivalent of a placebo. and the control group and re- During the period ofthe experiment,frequent measurements examinations are made tochart the course andeffectiveness of the treatment. of the measured The teacher pupilinteractions that were the essence "treatment" were consideredequivalent to a 5 mg.pill in a medical assumed that one pill experiment. Often, in theanalysis of results, it was much the same asanother; i.e., that,despite any personal was pretty administering the educa- misgivings of theinvestigator, any teacher tional "treatment" was prettymuch the same as anyother teacher. educational research is The crucial differencebetween medical and is in his pill, but the edu- that the biochemist knowsin great detail what pill, namely the teacher cator has little or noidea what is in his treatment pupil interaction. As aresult, the usefulnessof information coming studies for eithertheoretical or practicalapplica- from these evaluation obtained tions has been limited, to saythe least. The meager returns from such extended andexpensive projects haveled Smith and Meux (1962) to comment: consistent knowledge about in- The failure of suchstudies to yield a body of that more direct and primi- struction indicates thatperhaps they are premature; 7 '

tive analyses of teaching behavior areneeded as a preface to experimental and correlational studies (p. 2).

Research in Classroom Interaction Research often follows the developmentof technology. Just as the invention cf the microscope opened up a newworld previously unavailable to research scientists, theinvention and development oftechniques of recording (both auditory and visual)have opened up a new world to educators. Observational techniques have been usedby forerunners of this project in attempts tounderstand the classroom environment(Medley & Mitzel, 1963) and study the sociologyof small groups (Heyns & Lip- pit, 1954). Most projects have investigatedvariables relating to the emotional climate and control influences ofthe teacher on the students, an approach pioneered by the work ofAnderson (1939) in his study of dominative and socially integrative behaviorin the classroom. Some recent and more sophisticatedefforts in this direction are represented by the work of Flandero (1960, 1963), inwhich an involved observational technique of interaction analysis was used notonly to understand the ebb and flow of classroom processes,but also to train teachers to be aware of the influence they havein expanding or contracting the experience potential of theirstudents. These observational studies have madeimportant contributions in identifying variables for further investigation,providing preliminary data on problems such as measurement of judges'reliability, and developing statistical techniques to handle thedata. Nonetheless, there are some built in limitations toobservational techniques that need to be considered. Medley and Mitzel (1963) suggestedthat the validity of behavior measurements depends on thefulfillment of three conditions:

1. A representational sample of the behaviors tobe measured must be observed. 2. An accurate record of the observed behaviors mustbe obtained. 3. The records must be scored so as tofaithfully reflect differences in behavior (p. 250).

While there are serious questions related topoint 1 in most observational studies, it is point 2 that is of most concern.No matter how elabo- rate the observation or how manyjudges are involved, the essential limitation of the observation system is thatonly one type of behavior can be observed at anyparticular moment. Once that moment has passed, only that one segment of the totalbehavior remains as evidence of reality. Inability to re-create theessential raw data of the past situation presents an insuperable barrier tosophisticated analysis of the inter-

8 action and sequential development ofconcepts or the multiple interrelationships of variables operating simultaneouslyin the classroom. These problems can beovercome by the use of visual and auditory recordings, whichpreserve the experience so that its complexitiescan be analyzed more fully. The advantagesof this approach are shown in the increasing sophistication of thework of Smith and Meux (1962); Spaulding (1963); Taba, Levine, andElzey (1964); Hughes (1959); Perkins (1965); Withall (1960); andFlanders (1960). A number of investigators have triedto develop systems of cognitive behavior classification fora variety of purposes. One impressive effort is the taxonomy of educationobjectives described by Bloom (1956). This taxonomy was designed to enabletest constructors and curriculum builders to havesome common language of communication and topro- vide a structure through whichone could examine and evaluate education programs Thereare many similarities between this taxonomy and the present dassificationsystem in those categories directly relatedto the cognitive domain. A complex theoretical modelwhich also uses the concept of teacher student interaction andencompasses affective as well as cognitive domains has been proposed byRyans (1960, 1963). This model hasincor- porated computer terms suchas information processing and feedback as key concepts. Gage (1963a,b) has provideda useful summary of a variety ofpara- digms or models incurrent usage in the study of the teachingprocess. Which of these models ismost useful will be decided by theamount of research generated and the numberof fruitful hypotheses derived from each.

Research in Cogntive Processes In comparison to the development ofresearch in the areas of attitude and personality, which receivedgreat attention in psychological circles following World War II, interest in cognitivedevelopment andprocesses has languished. Complex thoughtprocesses have either been ignoredor dismissed as being adequately measuredby intelligence testscores. The idea that intelligence is whatan intelligence test measures has been followed in such practicalmatters as the definition of mentally retarded and intellectually gifted children. Thefew highly specialized laboratory studies on complex thoughtprocesses have not been much help in expanding concepts. Gallagher (1964a) noted four major differencesbetween the laboratory tasks and those encountered in themore normal learning environment of the individual (Table 1). Ina real life situation, a problem must be

9 TABLE 1 Differences between Laboratory Exercises and Reality in Use of Productive Thinkirga Laboratory Reality Problem selected for subject Problem must be chosen from many other stimuli competing for attention of the subject Extraneous stimuli reduced or systemati- Extraneous pressuresand distractions cally introduced randomly presented Problem chosen so it can be solved in aDifficult problems often extendfor short period of time weeks and months before solution Emotional involvement limited Emotionalinvolvementoftensevere, affecting self-status, livelihood, family, professional career, etc.

a From Gallagher, 1964b, p. 353. selected by the individual from among many, whereas in the laboratory, the problem has already been selected for the subject. In a normal setting, extraneous pressures and distractions impinge upon the person trying to solve complex problems, but in the laboratory, extraneous stimuli are reduced or systematically introduced. In reality, difficult problems demanding complex thoughts often extend over weeks or months, whereas the laboratory problem must be chosen in such a way that it can be solved within a short period of time. Finally, and perhaps more important, the experimental task is often aninadequate sample of the construct being investigated. Is a conditioned eye blink really isomorphic to the construct of anxiety? Is the inability to solve the water jar problem isomorphic to rigidity? Is the ability to find many uses for a brick really isomorphic to creativity? It is in this areathat results of laboratory studies should receive strictest scrutiny. The failure of the laboratory task to come to grips with more important constructs has been reflected in the comments of reviewers. Cor- man (198) commented, "Problem solving meant the behaviorsthat researchers, who say that they are studying problem solving, study" (p. 459). Duncan (1959) made the following summary statement:

The field of problem solving is poorly integrated. The reasons for this seem to be the use of a great variety of tasks to provide problems, the frequent useof unanalyzed and nondimensionalized variables, the lack of agreed upon taxonomy, and, to some extent, the failure to relate data to otherdata or theory (p. 426).

10 Two established trends bringhope for a more meaningful and systematic approach to complex thought processes.One is represented by the work of Piaget, the Swisspsychologist whose intensive case history studies of children's think'ng abilitieshave finally made a definite impact on American educationand psychology. The second trend is represented by the work of Guilford, whodeveloped a comprehensive picture of intellectual processes and operatio s.While these approaches differ quite strikingly, both provide a rathercomprehensive model of the cognitive processes and allow for thedevelopment of testable hypotheses which mayeventually build a picture of com lexthought development and operation. Guilford's early interest in creativity (1950a)seemed to play a large role in the development of his nowfamiliar structure of intellect (See Figure 1 on page 18). He dividesthe intellect into three major dimensions: content, operations, andproducts. The subcategories ofthese three dimensions are:

Content Operations Products Figural Cognition Units Symbolic Memory Relations Semantic Divergent Thinking Classes Behavioral Convergent Thinking Implications Evaluation Transformations Systems

Guilford's factor analytic researchconcentrated on the discovery of tasks that could measure all possiblecombinations of content, operations, and products. For example, the tasks listedbelow contain the same operations (cognition) and the same products(classes) but vary in content (Gallagher, 1964b, p. 178).

Which one does not belong?

A000 (Figural) Chair, Sofa, Table, Bed, Refrigerator (Semantic) 25, 49, 81, 14, 64 (Symbolic)

Guilford's model, which has undergoneprogressive revision with new evidence (1959, 1966), has the attractivefeature of including many

, 4.-=.7122L"

thought processes and operations thatdid not seem to be covered bythe usual intelligence test. For instance,intelligence tests have few items that call for divergent thinking,originality, flexibility, etc. Even so,it would seem likely that such a model wouldhave more theoretical interest than practical application exceptfor the fact that some of Guilford's tests have been used in educationalexperiments that obtained seemingly important results. One of the most influential of the studiesusing these new concepts was that conducted byGetzels and Jackson (1962), who distinguished between what they called "high creative" and"high IQ" adolescents. The high creative students were those whoperformed well on divergent thinking tasks but were not within the top 20 percent onIQ scores; the high IQ group had IQ scores in the top 20 percentbut lower scores for divergent thinking. Differences were found betweenthese groups in teacher rating, personality, self concept, and cognitivestyle. They stated, "One focuses on what is already discovered, theother focuses on what has yet to be known" (p. 14). Although this studyhas received virulently critical reviews because of a number of methodological errors(see, for example, Thorndike, 1963), it still contains manyrich ideas and concepts. There has been a continued flow of research on thecognitive style first noted by Getzels and Jackson (1962) and Torrance(1959). Reviews by Maccoby (1964) and Gallagher (1964b) havepointed out the confused and contradictory evidence available at this writing.Other observers have noted the similaritybetween the dichotomy of students noted above and similar divisions by Witkin, Dyk, Paterson,Good- enough, and Karp (1962)field independentand field dependent; Schachtel (1959)allocentric and autocentric; andMaslow (1959) growth and defense. Such research has revealedthe potential of the Guilford conceptual system for application to aneducational setting.

12 3 Objectives and Methods

This project sought to (a) identifyproductive thought processes in intellectually gifted children within the contextof classroom verbal activity at the junior high school level, and(b) assess relationships between these thought processes and certain variablesthat may relate to their operation in the classroom. Specific objectives were:

1. To apply a classification system to theclassroom verbal interaction of intellectually gifted children in order totabulate productive thought occurrences and to examine thefollowing propositions: (a) That children will differ significantlyin their patterns of classroom verbal behaviorin divergent, convergent, and evaluative thought processes; (b) That variations in the teacher's verbalactivities will relate significantly to variations in patterns of thought processesevi- denced in his students; (c) That verbal behavior in classes ofgifted children will show certain identifiable patterns and sequences,regardless of course content or type of school. 2. To relate manifestations of productivethinking in classroom verbal interaction of gifted children to other evidence, e.g., scores ontests of productive thinking. 3. To correlate measured differences inclassroom verbal activities of gifted children with measured differences inattitudes, self concepts, and other variables presumably related tointellectual productivity, in order to test the following hypotheses: (a) A low level of productive thinking inclassroom verbal activities will be associated with negative attitudestoward work, school, and society; (b) Frequency and quality of verbalizationin class discussion will relate significantly to self and self ideal concepts (e. g.,girls may talk less than boys because of differences inperceived role behavior).

13 Subjects The 176 subjects in the preset study were allacademically talented students enrolled at the junior and senior high school level in either a university laboratory school or apublic school. The selection of students for participation in this study was based on the fact that they happened to be attending the class sssions in social studies, science, and English which were tape recorded. The study covered five teachers, each of whom had two sections of students in the same subject area. The paired arrangement of the groups, e.g., BAKER-CHARLIE, wasused to indicate different sections taught by the same teacher. The BAKER-CHARLIE and HAT-IDEA groups represented almost the entire subfreshman class (combined 7th-8th grade) at University High School, University of Illinois. Students are admitted to thisschool on the basis ofperformance on the Lorge-Thordike Group Intelligence Test, Cooperative Achievement Tests, pastacademic records, and teacher recommendations. TheDAN-EASY groups reresented two sections of a freshman science class in the sameschool, withthe recordings taken one year later. The publicschool students in the study were in the FOX- GEORGE and JACK-KING groups. These studentshad been placed in accelerated science and English classes becauseof highscores on intelligence and achievement testsand evidence of past academic efficiency. Table 2 indicates by sex the means and standarddeviations of intelligence test scores andchronological age for each group. Each teacher had two sections of students insubstantially the same subjet area. The

TABLE 2 Chronological Age and Ability Level of Studentsby Group and S Age in Months Verbal IQ Nonver bat IQ N Mean SD Mean SD Mea SD

Boys BAKER-CHARLIE 23 145.87 4.35 130.96 9.12 128.09 10.69 DAN-EASY 15 166.53 8.20 130.13 7.19 128.80 10.79 FOX-GEORGE 29 159.86 3.61 131.17 6.87 134.59 10.28 HAT-IDEA 25 148.32 3.90 133.12 11.10 131.32 7.82 JACK-KING 26 175.54 3.54 129.73 9.93 132.54 9.85 Girls BAKER-CHARLIE 20 145.35 4.72 134,35 10.77 128.00 12.63 DAN-EASY 20 167.05 9.17 132.20 10.83 125.30 14.84 FOX-GEORGE 32 159.59 3.98 128.97 8.31 131.72 10.49 HAT-IDEA 17 149A7 4.08 127.24 9.07 122.88 8.45 JACK-KING 28 175.11 4.29 127.21 9.76 130.54 11.40

,. groups ranged in age from 12 to 15 years.The high standard deviation in CA in DAN-EASY was due to the fact that some members ofthe group had been previouslyaccelerated. The verbal and nonverbal IQ scores of the different groups were similar. The classes in this study can be considered groups ofgifted students, since the mean verbal IQ of all the groups would place these students in about the top 2 percent of the general population. Since these were group IQ test scores they were probably an underestimation of what the students would score on individual intelligence tests. Where both group and individual IQ test results were available, the group IQ test scores were almost always below thoseobtained on the individual measures, due to the lower ceiling on groupintelligence tests. As measured by these tests, there did not seem to be any meaningful differences in ability between the boys and the girls, with the inexpli- cable exception of the nonverbal IQ scores in the HAT-IDEA group in which the boys appeared to be superior. With this single exception, the sexes seemed reasonably well matched in termsof mental growth in both the verbal and nonverbal areas, both within the individual groups and across the total study population.

Data Collection The collection of data that would provide a faithful reproduction of classroom interaction presented a number of technical difficulties. The goal was to tape record 12 classes of intellectually superior children of junior high school age; the subjects covered were social studies, science, and English. In order to obtain a reasonable sample of classroom performance, it was decided to record five consecutive sessions of each class. Previous observations had indicated that one classroom session alone might not represent typical performance of either teacher orstudents, and, since many topics extend over more than one classsession, it was felt that assignments or discussion topics could only be observed by recording an entire school week. Table 3 indicates the classes that were tape recorded for analysis, a total of ten different classes with five teachers. The classes called BAKER and CHARLIE were tape recorded for five consecutive sessions in the fall and again for five consecutive sessions in the spring of the same school year in order to examine the performance of the same students in the same subject with the same teacher but with an interval of five months. Although the small nuMber of teachers and classrooms make any generalization extremely tentative,it was hoped that some interesting hypotheses might be generated. Another goal of the study was to look at the same students in different

15 TABLE 3 Classes Tape Recorded for Five Consecutive Sessions for Analysis of Productive Thinking Class Designation Subject Matter Level

BAKER-CHARLIE Social Studies University High School (Fall and Spring) Subfreshman Grades 7 and 8 Combined DAN-EASY Science University High School Freshman FOX-GEORGE Science Urbana High School Grade 9 HAT-IDEA Social Studies University High School Subfreshman Grades 7 and 8 Combined JACK-K1NG English Urbana High School Grade 10

subject matter classes. Many of the students in BAKER and CHARLIE were also in DAN and EASY with a different teacher and aswitch from social studies to science. In addition, many students recorded in their 9th grade science classes, FOX and GEORGE, were recorded one year later as members of the English classes, JACK andKING. Thus, similarities and differences in student patterns could be observed with different subject matter and teacher. Another source of comparison was the fact that groups HAT and IDEA covered the same subject matter as BAKER and CHARLIE at the same school at the same level but with another teacher. In addition, since each teacher had two different classes covering almost identical subject matter, it was possible to compare his performance with two different groups of students. The seating arrangement was kept the same as in the regular class meetings. Sound was picked up from three standing microphones, directed through a mixer, and finally into the tape recorder. Under ordinary circumstances, three members of the project staff were in the classroom during recording. One of the three persons operated the tape recorder and mixer, and the other two generally sat in the corners behind the students in positions where they could note who was speaking as well as any nonverbal information pertinent to the project. These observers tried to keep a running account of the sequence of speakers in order to identify the students on the tape recording, and they reproduced blackboard material or drew diagrams of experimental demon- strations, etc. They reported affect or emotional tone such as sarcasm,

16 wit, or intended humor. Twoobservers were used in order tocheck ac- curacy of studentidentification and pick upstudent by-play or unusual happenings in different parts ofthe room. Transforming the original taperecording into a tapescriptfor analysis was one of the most timeconsuming tasks in this project.The following sequence represents thetasks involved in thistransformation: 1. The tape recording andobserver notes were given to a secretarywho typed a rough copy. 2. The typed draft wasgiven to the first auditorwho was (when possible) one of the classroomobservers at the time ofthe recording. This auditor made modifications,which were often considerable, on the rough draft whilelistening to the original tape. 3. The modified draft wasthen given to a secondauditor (when pos- while fol- sible, the second observer),who again listened to the tape lowing the modified scriptwith observer notes. 4. The twice modifiedscript was given to aneditor, who prepared it in format for typing. 5. The secretary thentyped a ditto master and ranoff copies to be used for analysis. It is the consideredjudgment of the staff onthis project that until record- there is definite progress madein electronics, alternatives to tape ing need to be seriouslyconsidered in doing researchof the type in the present project. Evenunder excellent soundconditions young children do not make the best recordingsubjects. They are restless, movearound, and are often hard tounderstand. One alternativeconsidered by the project staff has been thepossibility of employing astenotypist to sit in the class and provide arecord of transactions. The taperecordings obtained from the class sessionsin the present study were notof sufficiently good quality to use forteaching purposes, despitethe cost of untold hours of staff listening timeand secretarial help.

Classification System It is no disservice tothose who have studied theclassroom environment, such as Aschner (1960),Flanders (1960), Smith(1960), Spaulding (1963), and Taba et al. (1964), tosuggest that we arestill in the beginning stages of understandingthe process of classroominteraction. In- deed, as Smith (1961)commented,

Our knowledge of the actof teaching as well as thatof taking instruction is justify from meager. Neitherof these acts has beeninvestigated sufficiently to .the act of teach- a scientificstandpoint, fundamental changesin teaching ..

17 ing has received far less attentionthan its central role in pedagogywould seem to require (pp. 93-94). One of the major prerequisites to aneffective analysis of teacher pupil interaction is some theoretical structure ormodel by means of which one can organizethe complex data. A good listof available models is given by Gage (1963b). At the timethe present research projectbegan there was much interest in thetheoretical contributions of Guilford (1956, 1959, 1966) and his structureof intellect. Theoretical aspectsof the project were highly influencedby his contributions, not only inthe classification system but also in thechoice of measuring instruments. Guilford's model of intellectualperformance was developed through a series of sequentialstudies employing factor analyticmethodology. From these he derived the wellknown model shown in Figure 1. The Aschner-Gallagher Classification System(Aschner et al., 1965), used in this project, was developed fromthis model. Of the three majordimen- sions of Guilford's model, thisstudy was most concerned with opera-

Units

Classes

Relations

Systems

Transformations

414%0.1°0Implications

1k11110.0110Evaluation Figural Convergent Production Symbolic Mvergent Production cb Semantic Memory .47.,4.4 Behavioral Cognition vsc, 01P'-

Figure 1. Theoretical Model for the CompleteStructure of Intellect (Guilford, 1959)

18 study dons. Thus, theprimary classificationcategories in the present memory (CM);convergent thinking(CT); divergent were cognitive routine (R). A brief thinking (DT); evaluativethinking (E1"); and description of each areafollows:

Cognitive memory representsthe simplereproduction of facts, 1. through such pro- formulae, or other itemsof remembered content (Although cesses asrecognition, rote memory,and selective recall. cognition and memoryin his model, they are Guilford separates differentiating them in combined here because ofthe difficulty in from our point of view, class interaction andbecause both represent, nonproductive thinkingoperations.) Examples ofcognitive memory performance can be seen inthe following:

T: What planet isclosest to the sun? Mercury, and it doesn't rotateeither. Bill: about T: What were someof the main pointscovered in our discussion mercantilism? attempt to keep Mary: One of the things welearned was that there was an a favorablebalance of trade. sixteenth President of theUnited T: Does anybodyremember who was the States? Bob: Abraham Lincoln. interchanges do notrequire the student to These teacher student direct integrate or associatefacts; the questions canbe handled by reference to the memorybank and selection ofthe appropriate re- information sponse from a storeof remembereditems. While factual higher thought processes,it is indispensable tothe development of moves intothe chal- is a sterile anduninteresting class which never lenge and excitementof more complexoperations. of given 2. Convergentthinking represents theanalysis and integration data. It leads to oneexpected end result or answer or remembered in- because of the tightlystructured frameworkthrough which the thinking fol- dividual must respond.Some examples of convergent low. would I have T: If I had 29 cents and gaveJohn 7 cents, how much money left? Bob: Twenty-two cents. in T: Can you sum upin one sentence what youthink was the main idea Paton's novel, Cry TheBeloved Country? Pete: That the problemof the blacks and thewhites in Africa can only be solved by brotherlylove; there is no other way. Thus, convergent thinking may be used in solving aproblem, summarizing a body of material, or establishing a logical sequence of ideas or premisesas, for example, in reportingthe way in which a machine works, ordescribing the sequence of steps necessary for passage of a bill throughCongress.

3. Divergent thinking represents intellectual operationswherein the individual is free to generate his own ideas within a data-poorsitua- don, or to take a new direction or perspective on agiven topic. Some examples of divergent thinking are: T: Suppose Spain had not been defeated when the Armada wasdestroyed in 1588 but that, instead, Spain had conquered England. Whatwould the world be like today if that had happened? Sam: Well, we would all be speaking Spanish. Peg: We might have fought a revolutionary war againstSpain instead of England. Tom: We might have a state religion in this country. These examples represent teacher stimulated divergent thinking but it need not always be teacher generated. In a discussion of the spoils systcm a student may generate the following: Pete: Well sure, the spoils system might be a good thing when a political party is getting started, but what about whenthere's no party systemlike in the United Nations? Here the student reveals his ability to take off from an established fact or facts and see further implications or unique associationsthat have not been requested, or perhaps even thought of, by the teacher. Instances of self initiated student behavior would also fall underthe general category of divergent thinking. 4. Evaluative thinking deals with matters of judgrnent, value, and choice, and is characterized by its judgmental quality. For example: T: What do you think of Captain Ahab as a heroic figure in Moby Dick? Bob: Well, he was sure brave, but I think he was kind of mean the way he drove the men just because he had this crazy notion of getting back at Moby Dick. T: Is it likely that we will have a hard winter? Mary: Well, I think that the pattern of high pressure areas suggests that we will. In the first example the student is asked to construct his own value dimension, what he considers heroic, and then to make a judgment as to where he would place CaptainAhab. In the second response,

20 the student is asked to give a speculative opinion or assessmentof probability. A third possibility, not illustrated here, involves entering a qualification or disagreement which modifies a prior judgment by another student, or stating a counter judgment which directly op- poses the statement of the previousspeaker. 5. Routine indudes procedural matters such as management of the classroom, structuring of class discussion, and approval or disapproval of an idea or person. Included here are the attitudinal dimensions of praise or censure of others or of self, as well as dimensions of structuring, such as prefatory remarks about what the speaker intends to say or do, or whathe expects someone else to say or do. Other characteristic occurrenceshumor, ordinary routine classroom management behaviors such as requests toclose the doorare included in this general category.

Secondary categories within the classification system were derived em- pirically by studying the classroom performance of students andteachers. Examples that seemed to differentiate certain behaviors from other types of behaviors were extracted, appropriatelylabeled, and classified under the prime categories. The final organization of thesecondary categories is presented in Table 4. The evolution of the secondary categories involved two major steps. During the initial stage, project members concentrated on detecting behavior which did not seem to fit into the established categories. These behaviors, if they seemed important enough, were made into new categories, given distinctive definitions, and placed under the appropriate primary category. This approach, while interesting and instructive to the staff in terms of how many intellectual variations could be obtained in a classroom setting, became unmanageable from the standpoint of obtaining consistent ratings by judges. Distinctions became too fine. It appeared, upon further consideration, that clear distinctions couldbe made between various cognitive performances without suchdistinctions having important psychological or sociological meaning. Thus, the second major step was to condense and eliminate a number of categories that did not seem useful for the final analysis. For example, several categories had been developed such as Cla (clarification of a previous statement by adding a bit of new information) and Clm(clarification of a previous statement by restating already given information). Since this fine distinction did not seem to have importantmeaning in pedagogy, the two categories were condensed into one general category (Cl). The final classification system was reduced by approximately25 percent from the high water mark of categoriesbefore the final analysis was carried out.

21 !

TABLE 4 Secondary Classification Categories

I.Routine (R) This category includes routine classroom procedural matters such as management of the classroom, the structuring of class discussion, and approval ordisapproval of the idea or the person. Management Mq Question: Requests or invitations to speak; calling for questions, as,"Any- body have a question?" Mp Procedure: Announcements or procedural instructions,given or requested for individuals or the group as a whole. Ma Aside: Incidental or parenthetical comment; gratuitous content. Mnc Nose Counting: Calling for or responding with a show of handsfor a tally or canvas. Mfb Feedback: Request for or response with signs from group as towhether or not the speaker's actions or remarks are clearlyunderstood. Mw Work: Nonverbal actions or seatwork going on in connection with current discussion or dass proceedings. X Response unclassifiable primarily due to technical recordingdifficulties. Structuring StsSelf Structuring: Conventional prefatory move to signal content and purpose of one's own next remarks or behavior. Sto Structuring Other (s): Engineering next speech or actions of other(s). Monitoring other's performance. Pump priming to keep discussion going on a point already before the class. Stf Future Structuring: Forecast of future activity, study, learning, etc., beyond this particular class session. Stc Class Structuring: Focusing class attention on point to be emphasized or taken up; laying groundwork for question or problem; probing, pushing, adding data for bogged down class (teacher only). Verdict Ver Verdict: (+ or ) Impersonal praise or reproach on quality of academic performance of individual or group. Verp Personal Verdict: (+ or ) Personal praise or reproach of individual (occasionally by T on whole class). Negative Verp generally on deportment. S Agreement: (+ or ) Acceptance or rejection of content; conceding a point; not permission giving or procedural. Agr Self Reference: Speakef's personal report or comment upon or about self. Often conventional device; cautionary tactic. Du Dunno: Explicit indication that one does not know. Mu Muddled: Speaker confused, mixed up, flustered. Hu Humor: Remark of evident witty, humorous, or comic intent; response (usually laughter) to same.

II.Cognitive Memory (CM) CM operations represent the simple reproduction of facts, formulas, and other items of remembered content through use of such processes as recognition, rote memory, and selective recall.

22 , MTTSIM 4",1 17; 7.4,7 r'ff r!.-7,:ns+11.7.^,caly703.filt.,

TABLE 4continued of a work ScrScribe: Giving a spoken orwritten spelling or exemplification or expression. Recapitulation literal reading from text, paper, ornotes in ReqQuoting: Rote recitation or hand. Rep Repetition: Literal ornearly verbatim restatementof something just said. of past extraclass occurrence. Rec Recounting: Narration discussed in current Rev Review: Recitationof material which occurred or was or past class session.

Clarification intelligible Chn Clarifying Meaning:Rendering a previous statement more either by (a) restating orrephrasing or (b) addinginformative details. Rendering a previous statement moreaccurate Clq Clarifying Qualification: entering an explicit either by (a) entering arider upon the remark or (b) correction.

Factual taken to be matters FsFact Stating: Requests forand recitations of items of fact. assertion of fact or other Fd Fact Detailing: Spinningout further a prior which factual items werementioned. statements (As, Exr) in in the form of a Fm Factual Monologue:Reporting of factual material monologue during whichverbal exchange is conventionallyexcluded.

III.Convergent Thinking (CT) and integration of Convergent thinking is thoughtoperation involving the analysis result because of the tightly struc- given or remembered data.It leads to one expected tured framework whichlimits it. Translation Tr Translation: Shift ofconceptual material from symbolic orfigural content to semantic, or vice versa.

Association of comparison; As Association: Involvinglikenesses and differences; degrees and relationships of direction,spatial position and/orclassification, etc. Explanation Exr Rational Explanation: Asking ortelling why X is the case; why Ycaused X, etc. Substantiating a claim orconclusion by citing evidence. bad, useful, important, Exv Value Explanation:Asking or telling why X is good, etc. Justifying a rating,viewpoint, or value basedjudgment by giving reasons why. how something is done, Exn Narrative Explanation:Step by step account of given outcome. how a mechanism works, orwhat led up to an event or

Conclusion general re- Gen Generalization: Integrationof prior remarks by slightly more formulation.

23 TABLE 4continued

Cons Summary Conclusion: Summary reformulation, often serial or enumera- tive, of material treated in discussion or reading. Conl Logical Conclusion: Calling for a deductively drawn implication from material presented.

IV.Evalutive Thinking (ET)

Evaluative thinking deals with matters of value rather than matters of fact and is :y characterized in verbal performance by its judgmental character. Unstructured Ura Unstructured Rating: A value judgment produced or requested on some idea or item in terms of a scale of values provided by the respondent. Uju Unstructured Judgment: A value judgment produced or requested on some idea or item wherein the value dimension has already been provided. Structured Svp Structured Probability: An estimate or speculative opinion is given or requested as to the likelihood of some occurrence or situation. Svc Structured Choice: Speaker calls for or declares his position as a choice between alternatives (not between yes or no answers). Qualification Qj Qualified Judgment: An offer or request for a rider or modification to a prior value judgment. Also, attempts to make more precise the value dimension discussed. Qc Counter Judgment: Speaker declares a directly opposed position with respect to value statement of a previous classroom speaker.

V:Divergent Thinking (DT) In a divergent thinking sequence, individuals are free to generate their own ideas independently within a data poor situation, often taking a new direction or perspective. El ElaborationStructured or free(s or f): Building upon a point already made; filling out or developing a point, but not shifting to a new point, often by concocting instances or examples. Ad Divergent Association (s or f): Constructing a relationship between ideas, casting the central idea into sharper and often unexpected perspective, by comparisons, analogies, etc. Imp Implication (s or f): Extrapolation beyond the given; projection from given datatypically by antecedentconsequent or hypothetical con- structionto new point (s) of possibility. Syn Synthesis: Spontaneous performance, tying in, integrating the currentcen- tral idea with an entirely new point or frame of reference. May bea variation or reversal of a previous conclusion. Double Paired Ratings: The complex nature of verbal classroom interaction often required the combination of more than one of the above described categories. Note.The full classification system and instructions are presented in Aschner, et al., 1965.

24 Figure 2 gives a sample of the flow chart used in the currentproject to classify thought processes. Each class session ofapproximately one hour produced 30 to 50 pages of double spaced manuscript,and each flowchart covered about one page of that transcript. The particular flowchart in Figure 2 covers partof a social studies class, HAT I. The subject under discussion was theresistance of natives

Tapescript Rater lc Date Pages Covered ...... Categories

Page Response K CM CON EV DIV Notes No. No, SPcaLer

t8v Pr Ce4;$1d ICI) i44241144464

/4 gi 1---. ,SfC,R .1!.. ASP Uvr2(2)

yki ct /qI 1/9 RokiDu A s Sis Ger 150 14-Rchect

/51 U ra CI pn

,/fgRokd Fm AO 153 T 1/441-

J54-r vevt Adf Isv(0.) nia, 534 a 1 55Ad/aArt-

Figure 2. Sample Page of Classification Flowchart (fromHAT 1 Session)

25 to colonization of this land. ResponseNo. 147 by Coin was a clarifying qualification (Clq) correcting a previous statement by theteacher. Re- sponse No. 148 included awide variety of teacher behavior: the teacher structured (Stc) the class as to what would be discussed next;asked for review (Rev) of information on an area previouslystudied; made a unique divergent association (Adf) of his own; made arating (Ura) on how well the class was doing on their assignment; andended by asking for a convergent association (As). Robert, in response No.149 to the teacher's question, stated that he did not know (Du), announcedwhat he was going to talk about (Sts), gave a convergent association(As), and followed with an explanation for his previously given association(Exr). In statement No. 151, the teacher gave an evaluativerating of his own (Ura) and clarified the previous statement made by Robert(Clm). In No. 152, Robert launched into a factual monologue (Fm),punctuated by a divergent association (Adf). The teacher expressed praisein a positive verdict on Robert's performance (Ver), gave a divergentassociation himself (Adf), repeated the association question in statement No.148 (Rev), and finally asked the students to make a decisionan evaluative choice between two alternatives that he presented (Svc). After becoming familiar with the category system it was possible to follow the tapescript flowchart and have a reasonably good idea what was being discussed and how, withoutlooking at the tapescript itself. To evaluate the reliability of the classification system it is necessary to list briefly the classification procedures:

1. A tapescript was given to two members of the project staff,who independently classified each teacher and student statement on a flow chart. 2. The two staff members then compared their ratingsand tried to reach a consensus on any classification discrepancies. In many instances this was relatively easy because one judge could seethat he had overlooked an obvious point. One of the unresolvedproblems was the establishment of a set in allthe judges which tended to result in bias, kading to over- or underrepresentation of certainclassifica- tion categories. The consensus meeting was extremely important to counteract such sets. 3.In those few cases where the judges could not reach agreement, the full project staff (five to seven persons) decided on the classification.

Consensus of individual judges would appear to stabilize the judg . ment to some extent, but there still remained a possibilitythat one judge with a strong personality or a particular point of view could sway the judgment of the larger group when discussing particularly difficult rat-

26 TABLE 5 Percentage of Agreement by Two Consensus Teams Classifying One Session of IDEA 3 Team One Team Two Category N Percentage N Percentage Cognitive Memory 126 62 132 58 Convergent Thinking 29 14 38 17 Divergent Thinking 20 10 17 7 Evaluative Thinking 29 14 42 18 Totals 204 100 229 100 Total Statements Rated-296 Total Agreement-231 Percentage of Agreement-78 ings. To illustrate the degree of consistency of the classification system, one class session, IDEA 3, was selectedbecause of the range of ideas present. Two teams, operating independently,classified the entire session; team judgments were then compared asshown in Table 5. The two teams gave essentially the same percentageallotment to the major categories. Team 1 rated 62 percent of the total responses as cognitive memory whereas Team 2 rated 58 percentof the items as cognitive memory; Team 1 rated 14 percent of the items convergent thinkingand Team 2 rated 17 percent of the items convergent thinking. It may be noted that the total number of ratings in the cognitive memory area differed between the two teams.Team 2 apparently divided the statements into finer subcategories than Team 1. An example of such a rating can be seen in the following teacher statement: T: All right. Now, does this presentif this is the case, then (several words inaudible) what we put on the board we should all consider tentative, that is, not finalbecause you probably gained this information from a single source and there are many reasons why asingle source of information may not be accurate. It may be your interpretation ofitthat--which another going to another source might clear up or it may be thatsimply that the particular book that you are using is not accurate. Now, let me ask, what kind of problem this could present, possibly, to the colonistsif the House of Burgessesif the members of the House of Burgesses wereelected by the Supreme Council in London. What kind of thing might you have? Leland? Consensus Team 1 saw in this statement structuring of theclass (Stc), rational explanation (Exr), and a request for implications (Imps). Team 2 agreed with all of the categories seen by Team 1but, in addition, saw a pedagogical rating (Urap)by the teacher. Thus, Team 2 had four

27 ratings for this one statement, whereas Team1 had only three. Similar minor differences would result in the differenttotals shown in Table 5. Of the 296 statements made by teachers andstudents, total agreement was obtained on 231, or 78 percent.For the purposes of this evaluation, only those statements on which there was complete agreement or agreement on two of three, three offour, or four of five categories, were labeled "agreement," and any instances whichdid not meet these cri- teria were labeled "no agreement." In a test of reliability of the primary categoriesfor the classification system, Hutchison (1963) compared theperformance of two judges over 1,037 statements and obtained over 90 percent agreement.These results give some assurance that the classification system,if used as described, can provide reasonableconsistency from one set of judges to another.

Measuring Instruments The choice of measuring instruments in a researchstudy is as crucial as any other singledecision. Since investigation of classroomperformance of students hasbeen relatively infrequent, there were few guide- lines. The few available studies suggestedthat the important factors influencing class performance were in the areasof attitude, personality development, and cognitive style; i. e., areasthat have not been noted for highly reliable or valid measuringinstruments. Nevertheless, for an exploratory experiment, it was decided tosacrifice some degree of pre- cision and reliability in order to approach asclosely as possible those variables that seemed pertinent. Thus, inaddition to IQ and achievement measures, various instruments wereused in an attempt to examine cognitive style, attitudes, and values. While it is well known that the IQ score is highlypredictive of the performance of the individual in an academic setting, itis not the only useful predictor. When a sample has a relatively narrow rangeof IQ scores, as in the gifted groups in the presentstudy, other factors assume more importance in the predictionof individual differences in classroom performance. Thus, it was necessary tolook for tests which do not correlate highly with IQ but might be related to classroomperformance. The Uses and Consequences tests used in the present study wereadap- tations of those proposed by Guilford (1956). In factor analyses, Guil- ford found performance on certain tests to load highly on divergent thinking, a factor which he considered closely related to creative abilities. Both the Uses and Consequences tests have been used in other factor analytic studies, in which they also seemed to load significantly on this divergent thinking factor(Kaya, 1960; McGuire, Hindman, King, & Jennings, 1961). In addition, other studies (Getzels & Jackson,

28 1962; Torrance, 1962) haveindicated that performance onthese tests does relate to academicachievement. The Uses Test. As originallydevised, the Uses test wascomposed of six questions; the student wasgiven ten minutes for eachquestion (Kett- ner, Guilford, &Christensen, 1959). In the presentproject, the time limit was reduced to fourminutes and the test to threequestions in order to fit the timeallotted in the total battery.Two sets of questions were devised following the basicform of the Uses test.Having been assured that there were no right answers,the subject was asked tolist as many uses for a givenobject or activity as he couldwithin the four minute period. instructions have an Since it has been shownthat changes in time and influence on test results(Christensen, Guilford, & Wilson,1957; Gold, 1963), the preciseinstructions for the test aregiven below: that we We often become soused to using certainarticles in specific ways also. For example, a shoe forget that they can be usedfor many other purposes be used to start a fire. could be used to drive anail or a magnifying glass can of the different On the following pages youwill have a chance to think up some ways that certainfamiliar articles could beused. obtained from the In the presentproject, three major scores were Uses test: given to three questions. 1.Fluency, or the totalnumber of responses 2. Breadth, or thetotal number of generalcategories used. (I. e., one would be "building agarage" in the generalclassifica- use for brick it tion, construction.Another use, "to hitsomeone," would classify would repre- as a weapon.The total numberof such classifications sent the breadthscore.) classification 3.Flexibility, a measure of thenumber of shifts from one to another whenthe subject's responses areread in the order listed.

The scoring system for theUses test in the presentstudy was developed on a trial and errorbasis and categories wererefined through staff discussion and successive tryouts.Two trained judgesrated 100 cases; the interjudge reliability score was.99. Thus, there was reason to assume that the scoring systemcould be used by trainedjudges. Table 6 shows the stability overtime of ideationalfluency (as measured by the Uses test) as apersonal characteristic in someof the present given the groups. In thefirst instance, one classof twenty-four students months obtained a test retest same test againafter a time lapse of two correlation of .77. With theintroduction of differentforms, the test retest correlation over aperiod of one year with53 subjects was .46. 29

....1.0"0.*... ) ^a TABLE 6 The Stability of Fluency as a Personal Characteristic Uses TestFluency Group N Time Lapse r Forms

ABLE 24 2 months .77 MA JACK-KING 53 1 year .46 MB

Reliability and stability coefficients were generally comparable to those obtained in other studies. While they did not reach thereliability level of instruments such as achivement tests, the characteristics measured could be expected to fluctuate from one time to anotherin the same individual, since they depend uponattitude and cognitive style.

The Uses questions in set A were:

1.List on the paper all of the uses you can think of for a brick. 2.List all of the things you can think of that mightbring comfort if you were hot. 3.List how many ways water can be made to work for you.

The questions for set B were:

1.List all of the uses you can think of for a newspaper. 2.List all of the things that might bring you comfort if you werecold. 3.List all the uses you can think of for a screwdriver. The Consequences Test. The Consequences test has receivedpromi- nent use in test batteries investigating productivethinking or creative abilities. In this case, three questions were used, the subject havingfour minutes to respond to each. The subjects were given thefollowing instructions:

Inventions such as the automobile and television oftenchange our lives in many ways aside from theiroriginal use. For example, the automobile has changed the dating habits of our youth and television hasbecome used as a type of baby sitter for our children. Itis fun to try and think about what effects might be produced by other future changes.

The first form of the Consequences test used thefollowing questions:

1. What would happen if pills were developedwhich would substitute for food?

30 2. What would happenif everyone were bornwith three fingers and no thumbs? 3. What would happenif the average temperaturefor this area rose 15° all year round? The second form of theConsequences test included thefollowing questions: I. What would happenif everyone in the worldsuddenly lost their hearing? 2. What would happen toIllinois if it rained six months outof the year? 3. What would happenif everyone lived to be about200 years old? The scoring standards forthe Consequences test were a matterof considerable concern to the projectstaff. Attempts were made todevise a quality score,but each was abandonedbecause of low interjudge reliability. Finally, on the basisof interjudge correlation of.90 and above, four scores were derived.

I.Fluency. The total number of responsesgiven. 2. Breadth. The totalnumber of general categoriesused. (These categories, similar to the Uses test, weredeveloped through successive trials and discussion of protocolsthat generally were notbeing used in the research analysis.) 3.Problem-solution. A ratio score wasobtained between number of answers andnumber of actual solutions. The typesof answers produced revealed that the childrenhad different styles ofresponding to the Consequences test.Some children concentrated onthe number of problems that would resultfrom a given condition, whereas other children concentrated on waysof solving these problems. 4. Personal-nonpersonal score.The subject was scored onwhether or not he was insertinghimself into the answers; e. g., "Iwould go hungry," as opposed to,"People all over the world would go hungry." Recent work by Guilford, Merrifield,and Cox (1961) confirms thatthe Uses and Consequences tests areclosely related to two componentsof divergent thinking, ideationalfluency and spontaneousflexibility. Breadth scores seem to possess highfactor loadings on spontaneousflexibility, while total response scoresweigh heavily on the general factorof ideational fluency. Kaya (1960), inher investigations of measuresof creativity, indicated that both theUses and Consequences testsloaded highly on a flexibility factor.

31 Semantic Differential. Since the book, The Measurementof Meaning, by Osgood, Suci, and Tannenbaum (1957) waspublished there has been continued interest in the research potentialof the semantic differential technique. in this technique, the subject is given aseries of concepts to which he must respond on a continuum ofadjective pairs: good-bad, heavy-light, or beautiful-ugly, etc. Osgood (1962), in his recent summary of crosscultural research on the semantic differential, found that whenfactor analyses were calculated on the results, similar factors were found acrosscultures. He also commented that there is no such thing as thesemantic differential, that each use depends upon the particular conceptsand set of adjective pairs used. This study used the same scale as aprevious research project by Pierce and Bowman (1960). A comparison ofhigh achieving and low achieving gifted secondary school studentsshowed significant differentiation between both high and low achievers,and girls and boys. Analysis of the use of this scale is given in theAnalysis section. Sentence Completion Test. Choosing anappropriate personality measure for preadolescentchildren is difficult, and the recent accumu- lation of negative evidence regarding the validityof projective techniques has tempered enthusiasm. On the other hand,the risk involved in using an objective scale is that the test constructorforces the subject to insert himself into the formal structureprovided by the examiner, which may result in an inaccurate portrait of the subject's ownworld. The senior author of this report had used sentencecompletiGn tests in clinical situations with children and was impressedwith the usefulness of combining objective and projective measures. The sentencecomple- tion method can:

1. Be administered to large groups of childrenin a reasonably short period of time. 2. Sample large areas of the child's self conceptand attitudes about the world. 3. Allow the child to express himself freely withoutrestricting him to a formal structure.

The sentence completion technique was originally developedby Payne (1928) and Tend ler (1930) as an adaptation of the wordassocia- tion method. The scale used in this study was adapted from the Rhode- Hildreth Sentence Completion Test (Rhode, 1957). By choosing appropriate sentence stems, an attempt was made to direct the child into specific areas such as self abilities, environment, or abstract concepts such as achievement, competition, and creativity.For example, the stem "my mother" or "my father" forces the respondent to comment in some way

32 on family members. In addition to specific content stems, a number of stems were included because of their open endedness, such as, "I wish that...," "Most of the time I...." The scoring system concentrated on whether the affect revealed in the responses was generally negative, positive, or neutral. For example, the response, "My mother is often cross and irritable," would be scored as negative affect; "My mother is a wonderful person," would be positive affect; and "My mother is 5 feet 4 inches tall," would be neutral. The scoring system was based on a seven point scale ranging from -1-1- in high positive affect to--in high negative affect. The judge read the entire protocol, identified the number of the item which referred to the particular characteristic, and then rated that characteristic. The individual's raw scores for the various items were used in the component analysis and multiple regression analysis. To investigate interjudge reliability, 20 protocols chosen at random from the experimental group were independently scored by two judges familiar with the scoring manual. As Table 7 indicates, this sample yielded a total of 460 categories for which some affect valence was assigned. Judges agreed on valence direction 93 percent of the time. They agreed on identical valence 73 percent of the time and disagreed by one valence step only 24 percent of the time. The judges disagreed by two or more valence steps on 15 of the 460 statements, or 3 percent of the time. These results were taken as establishing good scoring reliability for the test itself. In studies on the general validity and reliability of sentence completion measures completed by Rotter (1951) and by Rhode (1957), validity coefficients ranging from .61 to .82 were obtained (generally by using the scores of expert raters correlated with sentence completion scores). Hiler (1958, 1959), by using the sentence completion test as compared with an objectively scored attitude scale, found significant relationships

TABLE 7 Interscorer Valence Agreement for Sentence Completion Test Percentage Type of Agreement for 23 Categories° Percentage Range Two Judges Agreed on Valence Direction 426 92.6 75-100 Two Judges Agreed on Identical Valence 335 72.8 50-90 Two Judges Disagreed by One Valence Step 110 23.9 2050 Two Judges Disagreed by Two or More Valence Steps 15 3.2 0-10

From R. Smith, 1962, p. 35.

33 and was also able to distinguishbetween early terminators andlong term psychotherapy clients. Similarfavorable findingswere reported on test retest reliability, stability, andinterscorer reliability. Teacher Ratings. Teacherratings (Table 8), obtainedon all students involved in the project,were concentrated ori certain dimensionsre- lated to the underlyingtheoretical structure of theproject. The dimensions presumed most relatedto divergent thinking abilitieswere foresight (the ability tosee implications) and unique ideas. Thedimension described in the rating scaleas analysis would seem to bemore related to convergent abilities. Selfmotivation, self esteem, and sociabilitywere chosen as potentially relevantpersonality and attitudinal characteristics which could bereadily observed by the teacher. Previous experiencewith teacher rating scales(Gallagher, 1960) suggested that, in orderto obtain meaningful teacher ratings,the dimensions must be reasonablyspecific and descriptive; it isadvisable to avoid a middle pointon the scale; it is advisable to forcejudgment into one of the categories, rather thanallow gradations of judgment;and the number ofsteps in the rating scale should berelatively small since a large number appearsto have a spurious distinctiveness. A comparison of thestability of the two majorareas of the scale, cognitive and social characteristics, revealed substantialconsistency in ratings. Twenty-threechildren were rated bytwo different teachers within a time interval ofapproximately oneyear, and correlation between the ratings of the teacherswas computed as .72 for the cognitive abilities and .50 forsociability.

Analysis of Data The data analysiswas divided into three major sections: 1. A statistical descriptionof the classroom performanceof the various teachers and classes involvedin the study. 2. A correlationalanalysis relating studentclassroom performance measures to outside variables: academic,personality, and attitudinal. 3. Comparative analysesamong the various subgroups of the total student population, dividedby sex, age, cognitive style,IQ scores, etc.

Statistical Description.In order to present the actualclassroom performance of students andteachers as categorized by theclassification system used in the study,a tally, by category, was made of the performances of each student and teacher inthe five consecutive class sessions. Proportions of the primarythought categorieswere calculated for 34 Teacher Rating Scale on Personality and Attitudinal Characteristics of Students TABLE 8 I. Self MotivationDimensions ARarely self needsstarter. encouragement. Often thinks Level One periodsUsually of self reluctance motivated. or in- Has Level Two throughCan be roused occasionally Level Threemuch encourage- uninterested.Seems generally Does apathetic not grow and Level Four II. Foresight Generallyup extra showsprojects an on ability his own. to Usuallyquicklydifference. fromsees these.further implica- Bounces back Showssonaltendsment, occasional toor beby apathetic. a project flash of orfore- per- interest. Otherwise Veryclassroom.excited pedestrian. Seems passive. Does not or intrigued with benessseenext.always mental implications to lookInsights seesleaps for what them.forward. oftenand is aAlmost seemreadi-coming to performance.slumpsdinarytions more often than the or- student.into unimaginative Sometimes tenprisesight. uninspired. to This the isteacher. generally More a sur- of- sequencesingthink discussed. of relatedimplications to issue or be-con- III. Self Esteem Rarely,orSupremely held if ever,back seems by self bothered doubt. self confident. consideredSometimespected,under since stress.a shows stable he Usually individual.selfis generally doubt unex- favorablebrightenstiveSeems view often of circumstances.up self. to under express Occasionally praise nega- or ration.ofvealattitudeShows great this consistently towards conceitby giving orself. impressionself May admi- re- negative IV. Unique Ideas fromraisedgoodupCan with be producing.quality in counteda class. unique to Hardmany on answer to to issuescome stop of placedisappointsfreshUsually ideas. comes up with some ideas, but sometimes with common- fresherally.ly.Shows Routineideas, occasional but performance, not consistent- flashes gen- of class,whatnovelSeldom teacher, idea.can or be Usuallynever or obtained text. produces gives from back a V. Analysis lutions.getheranalysis;Fxcels evidence inAn efficiently ideal leading detective. draws to so-to- tasks demanding ineventasks,Good productive performance, routineon analysis direction.in others. in but certain leans Un- long.lyze.Is formance.sometimes Usually Does stirredspotty not toin persist ana-per- or possible.sis.Seemssituations repelled involving by questions analy- Avoids them whenever Ui VI. Sociability ers.wayswithSocially peerseen popular. ingroup. company VeryAlmost of active oth- al- choice.seemsGetsCanties along towork prefer well individually, group with others.activi- but (work, social) if given time.ferwheninterests. beingOften necessary. Interacts alone alone, Seemsmuch pursuingwith toofothers pre-the own bywithit himself; Athat others. loner. way. seemingly Comes Spends and preferslittle goes time

--asysaystmagy teacher questions, teacher answers, boys' statements, and girls' statements for each class group. These proportions were used to make comparisons between different class sessions and different groups. In order to provide a rough comparability of individual classroom performance across class groups, the total of each major thought category for each individual was divided by the total class performancein that category. For example, if Sam had 10 statements in the divergent thinking category and the total generated by his class group was 100 statements in that period, Sam would receive an index of 10/100 or 010. This adjusted classroom score would be comparable to that of a student in another class who made 20 divergent statements out of a total of 200. These adjusted classroom indices were used as the basic classroom performance data in later component analyses. Chi squares were computed to indicate differences from expected or norm performance. Correlational. To obtain information on interrelationships between classroom performance and numerous academic, intellectual, and personality variables, the adjusted classroom indices, cognitive abilities test data, personality and attitude dimensions, and teachers' ratings were placed in a component analysis using the varimax criterion for analytic rotation (Kaiser, 1958, 1962). The semantic differential concept scores were obtained by summing the scores over the fourteen adjective pairs. The teacher ratings were reduced from six to two based on an initial factor analysis which indicated that five of these six rating scales clustered into a general cognition factor. An analysis of variance was induded to identify those variables that appeared to have the most influence on student classroom performance. In other correlational analyses, traditional parametric and nonparametric measures were used as appropriate. Comparative Analysis. In order to make a comparative analysis of classroom performance among recognizable subgroups of the total population, the performance data were analyzed and reviewed by sex, age, and cognitive style. The fundamental tool for this analysis was the t test, or chi square where a nonparametric test seemed appropriate.

36 4 Results

Results were organized to present data related to the specificproject objectives and to questions that emerged as the project progressed. The major questions asked were: 1. What were the proportions of the various thought processes expressed by teacher and by students? 2. What were the relationships between classroomperformance and outside variables? 3. What was the relationship between the cognitiveperformance of teacher and students? 4. Was the classroom cognitive style of gifted childrenconsistent over changes of time, subject matter, and teacher? 5. Was cognitive expression of teacher and student consistentwithin a given class section? 6. What relative 'influence did the group, class section (time ofday?), and day of week have on variance of verbalized productive thought in the classroom? 7. Was sex a relevant variable on classroom expressivenessand other variables? 8. Were there differences between cognitive style subgroups? 9. Were there differences between expressive and nonexpressive students?

Proportion of Various Thought Processes Expressed by Teacherand by Students Graphs were constructed by taking the total responses for a class session and dividing them into the four primary categories:cognitive memory, divergent thinking, convergentthinking, or evaluative thinking. Responses in the category termed routine were considered irrele- vant and not used in constructing these graphs.Distinction was made between teacher questions and teacher statements. Thc questions teachers ask were considered indicative of the typeof thought processes asked for on the part of the students. On the other hand, teacher statements appeared to represent a type of personal teaching style.

37 100 CHARLIE 100 GO 913 00 10 70 70 & 60 & 50 2 50 40 I a. 30 0.30 20 20 10 10 0 0 1 2 3 4 2 3 4 Social Studies SassionFall Soolal Studies SsionFall BAKER CHARLIE 100 100 00 00 110 110 70 70 & so GO

50 SO ip 30 1. 304° 20 20 10 10 0 0 1 2 3 4 1 2 2 4 Socil Studies SassionSpring Social Studies SassionSpring

Cognitive Memory MI Divergent122 Convergent Evaluative EJ

Figure 3. Proportion of Thought Processes Asked for by the Teacher

In reporting student behavior, all of thestatements were counted whether they were questions oranswers. It was noted in the recordings and observations of the classroom that students often seemedto be asking questions when, in fact, theywere giving answers. For instance, if the teacher asked, "Whowas one of the foremost leaders of settlers in the Carolinas?" a student might respond, "Wasn'tone of them Sir Walter Raleigh?" Early in the project, Dr. Aschner classified thesetentatively given student responses asquanswers. Since much of the question asking behavior of the studentswas of this nature, student questions and statements were combined into one graph. Figures 3 and 4 show the proportions of the four thoughtprocesses in five consecutive sessions of the BAKER-CHARLIE social studies class, recorded first in the fall of theyear and again the following spring. Figure 3 shows teacher variation fromone class session to another and from one group to another. For example, in BAKER-Spring,the teacher asked for cognitivememory responses more than 70 percent

38 BAKER CHARLIE 100 so ao

713 & 60 50

4400 40 30 30 20 lo 10

1 2 3 4 5 2 3 4 Social Studies SessionFall Social Studies SessionFall

CHARLIE 100 so $o

& SO a 40" 30 29 10

1 2 3 4 2 3 4 Social Studies SessionSpring Social Studies SessionSpring

Cognitive Memory IN Divergent 122 Convergent Evaluative C:1

Figure 4. Proportion of Thought Processes Given by. Boys of the time in each session, whereas in CHARLIE-Spring, the range of such questions was from 40 percent to over 70 percent. Despitethe fact that this teacher was considered skilled in stimulating good thinking, approximately 50 percent of the teacher questions were of the cognitive memory type. This seems an almost inescapable average for a teacher who deals in detailed factual information. Even when problem solving or evaluative conclusions are the goals, it is necessary for the students to have a broad base of information. It is likely that practically all teachers at this grade level produce a baseline of approximately 50 percent memory questions. In some class sessions certain kinds of thought processes were not requested at all by the teacher. For example, in BAKER-Fall 4, there were no requests for evaluative thinking ordivergent thinking. In BAKER-Spring 3, there were no requests for divergent thinking, and in BAKER-Spring 5, no requests for evaluative thinking. This again underlines the importance of obtaining a large sample of observations on a given teacher, lest the conclusionbe made that this teacher never

39 calls for some kinds of intellectual responses. Resultssuggest that, on given days, a teacher does not ask for certain kinds of intellectual behavior. Another pattern was noted in the BAKER-Fall series. Divergent thinking occupied an appreciable proportion of class time in the first three sessions but did notappear at all in sessions 4 and 5. In the BAKER-CHARLIE Spring sessions, however, divergent thinking seemed to be spread rather thinly and equallyover the five sessions. Thus, a particular teacher's performance is relatedto the general flow and purpose of classroom operation ata given time. A comparison of student performance (Figure 4) with teacherre- quests in these same classes (Figure 3) indicates the dependence of the expressed student thought processesupon the type of teacher questions asked. Thus, in the CHARLIE-Fall sessions 1,2, and 3, extensive teacher requests for evaluative and divergent thinking producedthe highest percentage of productiveresponses on the part of the boys. Al- though all data were calculated separately for boys and girls,both had essentially the same proportion ofresponses and reporting both would be repetitious. In the area of divergent thinking, itwas obvious that a small percentage of teacher requests can bring fortha much larger percentage of response from the students. Open ended questions stimulateda great many student statements. For example, the teacher question, "What would have happened if the Spanish Armada had been victoriousover the English?" brought fortha veritable snowstorm of student responses along many dimensions. The figures also indicateddifferences between groups. CHARLIE and BAKER groups, one of which met in the morning and one in the afternoon, appearedto differ in proportions of thought processes despite the fact that the subjectmatter was nearly identical. Figure 5 presents the five consecutive class sessions oftwo teachers in two different contentareas of science. The DAN and EASY sections were grade 9 classes in biology, and the FOX and GEORGE serieswere grade 10 classes in physics. Proportions of thoughtprocesses requested by the teachers confirmed general expectations.Teachers in these science courses tended to concentrate on cognitive memory andconver- gent thinking rather than divergent and evaluative thinking. There was less teacher variation from one class to another than in thecase of the social studies teacher. Therewere, a priori, some expectations that science courses would present different proportions of thethought processes than English or social studies, and these expectations were generally realized. As previously noted, there were instances in various class sessions in

40 DAN EASY 108 too 10 oo

GO SO 70 70 ea 58 50 40 30 20 20 10 10 0 1 2 3 4 1 2 $ 4 Science (Biology) Sessions Science (Biology) Sessions FOX GEORGE 100* so $o

I 6° 50 40 30 24 10 0 3 4 5 1 2 3 Selene. (Physics) Sessions Science (Physics) Sessions

ICognitive MemoryEa Divergent El2: Convergent En Eve !until/ft El

Figure 5. Proportion of Thought Processes Asked for by theTeacher which certain kinds of thought operations werecompletely shut out. No questions related to evaluative or divergentthinking in FOX 4, and the same situation prevailed in GEORGE 2 and DANS. There were no requests for divergent thinking in GEORGE5, and so it went. In the FOX and GEORGE series, divergent andevaluative thinking made up less than 10 percent of the total teacherquestions in any one class session. The percentageof requests for these two thought processes was somewhat higher in DAN-EASY,but at no point did the combined percentage for the two areas go above 20 percent.On the other hand, convergent thinking often reached 50 percent or more. In Figure 6, student performance in these courses canbe observed to follow rather closely the requests for certain thought processesby the teacher shown in Figure 5. A relatively smallnumber of divergent thinking questions brought forth a substantialproportion of student responses in this area. The consistentlyhigh level of convergent thinking in these classes (as compared to social studies) seemed toreflect the demands of the teacher. Since teachers in the sciencesections varied

41 less in performance from day to day, student performance also seemed to vary less. In comparing the styles of the two teachers, the DAN-EASY instructor asked more cognitive memory questions, while the teacher in the FOX-GEORGE group asked for more convergent thinking responses. Figures 7 and 8 present the classroom periormance of the final four groups in the study. One teacher taught both the HAT and IDEA sections of social studies; another teacher taught both the JACK and KING sections of English. The differences in teaching style are reflected in the different proportions of thought processes. The social studies teacher asked for a greater degree of evaluation than did either the English teacher or, apparently, the social studies teacher shown in Figure 3. These graphs also suggest that the type of questions asked by the teacher depends partly on the structure of the group. There were clear differences between types of questions asked in the HAT and IDEA sections despite the fact that the same material was covered on the same days by the teacher. With both teachers, considerable varia-

DAN EASY 100 go

Or, 60 & 50 50 30 1 40 40 a. 30 30 20 20 10 10 0 0 2 3 4 1 2 3 4 Science (Biology) Sessions Science (Biology) Sessions

FOX GEORGE 100 100 .1. 90 90 7.4 $0 90 70 70 ao 40 & 60 5° 5° 140 4° a. O. 33 30 20 20 10 10 0 0 2 3 4 2 3 4 Science (Physics) Sessions Science (Physics)Sessions

Cognitive Memory Divergent1221 Convergent Evaluative EJ

Figure 6. Proportion of Thought Processes Given by Boys

42 class. No one tion was noted from onesession to another in the same class session could beconsidered typical. questions In the JACK andKING English sections,the teacher's predominantly cognitive memoryand convergentthinking. Only were thinking occur in more in KING 2 did eitherevaluative or divergent divergent thinking were than 15 percent of theinquiries. Requests for infrequent; such questionsdid not seem to be a partof this teacher's style. (Figure 8) followed As might be expected,student performance Since very little rather closely the patternrequested by the teacher. teacher, little stu- divergent thinking wasencouraged by the English dent divergent thinkingoccurred in JACK andKING sections. How- proportion of convergentthinking was noted. in response ever, a high and IDEA to more requestsfor evaluative judgmentin both the HAT produced a substantialproportion of evaluative groups, the students 5 and 6). responses. This isin contrast to thescience groups (Figures In summary, much of anyteacher's questioningis in the cognitive

HAT 100 100 10 10 10 10 70 70 & so 40 ac50 gg I40 Z40 30 30 D. 20 20 10 10 0 2 3 4 3 Soziel Studies Sessions Social Studios Sessions KING JACK 100 100 10

$O 10 70 70 & 50

40 45°5 Z 30 20 20 10 10 0 0 2 1 2 3 English Sessions English Sessions Cognitive MemoryMI Divergent122 ConvergentSiA EveluMive

by the Teacher Figure 7. Proportion ofThought Processes Asked for 43 memory area, and convergentthinking seems likely to be the nextlarg- est proportion. Whether or notdivergent thinking appears instudent performance depends on theteacher's own philosophy andstyle of teaching. A very respectableclassroom (in terms of cognitiveperformance) can be operated withoutdivergent thinking being requested at all. The same could not be saidabout cognitive memory or convergent thinking. These figures suggest that differentsubject matter areas seem topull different proportions of thought processes,although teachers still show individual styles within a given subject. While the preceding figures give anoverall picture of the distribu- tion of responses in the major categories, animpressive range of individual differences between teachers,within teachers, and between classes is shown in a breakdown of the secondarycategories. Table 9 indicates secondary category questionsasked by teachers in the five day series for each of the 12 dass groups.In the area of cognitive memory, requests for recapitulation or pure memoryitems ranged

IDEA HAT 100 100 so so $o To so so so 50 40 40 a.30 30 20 20 10 10 0 1 2 3 4 $ 2 3 4 Social Studies Sessions Social Studies Sessions KING JACK ioo 100 $o 00 60 se 70 70 60

SO 40,

30 30 20

10 10 0 3 2 3 2 English Sessions English Sessions Cognitive Memory III Divergent71 Convergent Evaluative

Figure 8. Proportion of Thought ProcessesGiven by Boys

44 section. Facts re- from 30 teacher questionsin HAT to 90 in the FOX quested vary from 31 inBAKER-Fall to 153 in DAN. Since it is tempting toascribe such differencesin performance to teachers of the differences in subject areas,it is important to compare thinking questions were same generalsubjects. Many more convergent requested in FOX-GEORGEthan in DAN-EASY, yetall were science sections. Many morefactual questions were askedin DAN-EASY than influenced more by in FOX-GEORGE. Theseresults were apparently differences in teacher stylethan by the pressures ofthe subject matter. Similarly, the teacher in thesocial studies sections ofHAT-IDEA asked thinking than did the for considerably fewerfacts and less convergent teacher in the social studiessections of BAKER-CHARLIE,although both covered the samesubject matter and almostthe same topics. In convergent thinking,the range in numbersof teacher questions seeking conclusion statementsextended from a low of 1in CHARLIE- Fall to a high of 43 inGEORGE. In this respect,it is of interest to com- with pare the numberof conclusion statementsmade by the teacher the number asked of thestudents. In most classsections, teacher con- questions by a ratio clusion statements outweighedteacher conclusion ranging from 3 to 1 to 10 to 1.Only one teacherrequested more conclusions from the students thanhe gave himself. The goal of this project was not toarrive at some theoreticalideal of teaching style, but rather todescribe what actuallyhappens. If, for example, the teacher had acommitment to the discoverymethod, in their which teachers encourage thestudents to reach condusions on to own, then aratio of 5 to 1 or 6 to 1 ofteacher condusion statements alarm. If, on teacher conclusion questionswould be viewed with some the other hand, the goal wereto presentconcisely and clearly the most important point in thediscussion, such a ratiomight not be looked on with disfavor. In the social studiessections, the translation(Tr) column under involved map reading; inthe science sections,it convergent thinking mathematics to the involved moving from thesymbolic language of reasoning to justify semantic area. Requests forexplanation or logical positions or conclusionsvaried from 10 in IDEA to92 in FOX. divergent thinking and As previously noted,teacher questions on evaluative thinking werefewer in comparison tocognitive memory and convergent thinking.In the elaborationsubcategory surprisingly few stressing extrapo- teacher questions wereasked. Divergent implications, lation or projection of anidea from given data,received only slightly 11 in more attention.The range was from0 requests in KING to CHARLIE-Fall. As was truewith other divergentquestions, one teach- from the students. er question canbring forth a volume of responses 45

gat_ ib Secondary Category Classifications of Teacher Questions Teacher Questions TABLE 9 and Statements Groups Routine Cognitive Memory ConvergentThinking DivergentThinking EvaluativeThinking All BAKER-FALL St 1 0 ± Mu0 0 Hu 0 Scr 0 77Re Clm27 Clq7 31F 11Tr &0 22 Ex29 Con Total 2 53 El2 Ad 4 Imp Syn Total8 0 1415 U4 1614 S 2Q Total 2218 Total 2252.32 BAKER-SPRINGCHARLIE-SPRINGCHARLIE-FALL 021 0 01 01 0 40 724049 483130 645 791448 40 211418 252323 2I3 424558 202 794 11 59 0 2014 340 7 0 10 7 224275 EASYDAN 31 0 00 00 00 01 4664 17 7 10 0 153141 50 10 48 292852 923231 41 7 195 7571 00 1413 3 64 0 2017 5 305 11 6 00 1611 9 302349389 FOXHATGEORGE 232 00 00 00 0 620 589030 302534 653 4178 18 1 1817 8114 43 9 162 42 10 48 246 0 1310 82 21II 7 0 29 9 199358 JACKI DKING F A 31 00 00 0 00 201 485930 4116 13 43 435136 02 413714 181410 2825 4 847930 03 520 301 00 11 21 13 96 1613 9 0 222519 231224188 TABLE 9continued Groups Routine Cognitive Memory Teacher Statements ConvergentThinking DivergentThinking EvaluativeThink:ng BAKER-FALL 126 St 23 41+ Mu Hu 7 6 Scr 5 107Re Clm Clq98 33 F27 Tr As6 20 Ex25 Con49 Total 100 16El Ad Imp Syn9 Total 6 0 31 58 U 12 S 7Q Total 77 Total 681All CHARLIE-FALLCHARLIE-SPRINGBAKER-SPRING 224 154183 246336 635955 231110 181213 11 48 174119176 112132120 766555 505478 1519 2 242830 343633 516361 115138143 13 68 261412 11 94 0 332941 716955 251217 10 35 104 7093 1,066 871992 DANFOXEASY 184177167 3511 6 22 06 151811 25 41 1417 6 290263280 113 6876 397855 168121 57 241812 346013 329383 605914 231188 83 448 2012 8 37 0 351915 125 2898 301019 2 130144 40 1,2101,145 857 HATGEORGE 268188139 271812 3614 3 12 13 38 93 421 208 6588 4559 2210 1683 36 8 181512 19 79 353317 616784 241 11 98 541 10 201610 482426 376 0 513033 513461835 IDEAKINGJACK 276236 2410 1214 51 1820 1 232136 161120 7965 9 4853 9 03I 3761 6455 7766 203161 10 6 79 40 0 2115 109 96 2010 23 118122 1,198 954

.14 SN

In Table 9, the teacher statements arecategorized reflecting intellectual and management expressions not generallydesigned to elicit student comments. While types of questions askedindicate the goals of the lesson, statements seem to reveal moreof the teacher's personal style. Under Routine category in Table 9, a wide range of differences among teachers in theexpression of negative and positive verdicts can be noted. Verdicts represented teacher approval or disapproval of student behavior or performance. The ratio of positive to negativeverdicts was of interest, as well as theabsolute number of verdict statz-ments. The most favorable ratio was found in CHARLIE-Spring wherethe teacher gave 63 positive verdicts as opposed to 24 negativeverdicts over the five sessions. At the other extreme was the FOX sectionin which the teacher gave no positive verdicts and expressed 35 negative ones. Group differences influenced the verdict ratio, as can be seen indiffer- ences between sections which had the sameteacher. For example, in DAN the positive to negative verdict ratio was about 2 to 1, whereas in EASY, the afternoon section with the same teacher, the ratio was 1 to 1 with fewer verdicts of either kind. The ratio was fairly even in HAT, but in IDEA there was a 2 to 1 positive to negative margin.Simi- lar differences between sections with the same teacher were found in the JACK-KING group. It is of interest to compare the positive to negative verdict ratio in the boys' statements. (The girls, as seen in Table 10, do not often give verdicts of either sort.) The ratio of the boys' verdicts seemed highly related to the ratio of the teacher verdicts; a Spearman rho correlation of .82 was found in the verdict ratios of teachers to verdict ratios of boys. No simple causal relationship need be inferred. It is likely that some complex affective balance was establishedbetween teacher and students in each classroom. In the category Humor, the range extended from one such teacher statement in DAN to 38 in the GEORGE section. As with otherfigures, these numbers obscure complex behavior. Humor can be either directly related to the classroom material, or it can be off beat, such as a humorous story which appears to derail the classroomdiscussion rather than further it. As an example of another type of humor, the casual intro- ductory statement by the teacher, "Now that we no longer have day- light saving time, it was easy to get up this morning," would be classified as humor, as would be the student response, "It was?"Such humor seemed to create a more relaxed atmosphere without interfering in the academic operation of the classroom. Another type of humor, suggesting that all was not as it should be in the teacher student relationship, was noted, for example, in the statement:

48 T: I will now hook up the apparatus and we'll seewhat happens. John: If you're in your usual form, nothingwill happen.

Interchange of this nature, particularly ifthe sentiments are supported by other students in the class, carry abite quite different from the casual quip or play on words that occursin most classes of gifted children. In dispensing of facts, as shownin Table 9, the range of teacher statements extended from a lowof 9 factual statements in the IDEAsection to a high of 168 in the EASYsection. This marked differenceseemed to reflect that information dispensingcharacterized the DAN-EASY sections, in contrast to the HAT-IDEAclasses where less time was spent on specific factsand more with the abstract conceptof colonization. Variation in convergent thinking amongteachers extended over each of the subcategories.Translation activities appeared morefre- quently in the science classes thanin English and social studies.Expla- nation occurred most often inscience and least often in socialstudies. In sections taught by the sameteacher there was more consistency in teacher statements than in teacherquestions, supporting the notion that personal teaching styles areinvolved. In divergent thinking, socialstudies teachers did not appearstrik- ingly different from science or Englishteachers. The English teacher gave a representativenumber of divergent thinking responsesalthough she asked for little divergent thinkingfrom her classes. Elaboration, or the development of examples,appeared more often in socialstudies and English sections, as might beexpected. The drawing of implications or extension of ideas were clearly notexclusive to any subject area. It was a surprise to see that one ofthe science teachers showed greater expression of evaluative thinking thaneither of the social studies teachers or the English teacher. Thiscould apparently be accounted for on the basis of personal style. Forexample, the teacher would say, "So bacteria can be parasitesright on human beings. Itis not very pleasant to thinkabout but actually this can be true."This statement represents a personalevaluation, not necessarily essential tothe content. All in all, the teacher statementsin the secondary categories were probably betterreflections of the wide range ofindividual differences amongteachers, or within a given teacher'sperformance from one time toanother, than those observed in theprimary categories. Table 10 shows the total performancein the secondary categories by the boys and the girls in all sessions.Comparing total verdicts, it can be seen that in almost all sections, boysoutproduced the girls. The boys gave 127 negative verdicts and 25positive ones. In contrast, the girls gave 49 negative and 9positive verdicts. The verdictratio remained practically the same for either sex,i.e., 5 to 1 negative to posi-

49 4M TABLE 10 Secondary Category Classification of Student Boys' Statements Statements 11 Groups Ver Ver Du Fe Routine -1- Cognitive Memory ConvergentThinking DivergentThinking EvaluativeThinking All BAKER-FALLBAKER-SPR1NG 2520 St 24 30 11Mu19 Hu 11. 4 Scr 03 7873Re 28am15 3314 Clq 5746F Tr As73 3416 Ex Con7559 Total 1110 127 88 I El Ad Imp11 Syn Total4 29 58 28 21 01 91 61 2619 U 3026 S 13 8Q Total 6953 Total 505430 DANCHARLIE-SPRINGCHARLIE-FALL 62 39 054 2 131015 4819 10 10 122 65 5633 3966 112144 98 21 1 2015 5651 24 6 121 73 12 5 6227 3967 10 114 99 23 6 2111 13 5 5722 816509 FOX.EASY 60 80 14 8 03 29 6 42 09 32 180 5339 562514 106 1710 135146 29 24 852829 151 3028 58 87 323 68 310 392621 23 89 0 653530 15 33 2215 7 11 00 481810 1,061 390305 IDEAHATGEORGE 574291 41 07 10 12 121024 5514 5 22 112 4041 302967 109 4771 111 9642 57 3 252238 16346 221691 349 93 16 0 3027 31 9 10 7836 1416 2611 64 4631 1,044 477 KINGJACK 3416 37 5 20 53 21 5 02 8248 5128 5634 5527 I 021 5257 383233 2943 134118 76 017 18 4 18 03 02 45 75 1421 7 1618 331 263142 515315453 .Ctryjr2r.oyIMIVM.P157,C, TABLE 10continued Girls' Statements Evaluative Croups St Ver Ver Da Sc Routine Scr Cognitive Memory Clq ConvergentThinking Ex Con Totai El Ad Imp SynDivergent TotalThinking U ThinkingS Q Total Total All BAKER-SPRINGBAKER-FALL 10 -- 01 + 0 Mu Hu 03 03 10 48Re 1 am 8 40 29F 3 Tr As51 1216 2721 73 4548 53 1113 2010 1 2737 1216 12 7 5 1 2528 219184 CHARLIE-FALL 28 2 11 21 79 32 0 4073 1013'0 1514 41 0 3 16 2331 13 2 5652 71 1418 5217 20 4171 16 3 11 7 6 1 3311 319302 EASYDANCHARLIE-SPRING 33 35 0 0 54 0 040 2347 67 67 8774 254 1710 14 5 22 3819 0 96 30 0 12 6 43 24 01 7 168195 FOXGEORGE 315541. 23 80 051 2430 8 3631. 00 124 4193 225446 937117 126 5895 10 7 404411 607617 473511 154165 41 30 1412 6 73 00 2419 9 15 59 12 92 650 2927 7 261685641 JIDEA AHAT C Z . 53 8 01 0 12 5 20 61 3931 2318 2415 2960 20 3212 2719 2914 5380 08 901 71 00 24 1 24 3 1315 1 53 44 7 203331275 VI ICING ' 14 14 0 4 11 0 40 15 31 31 0 30 32 28 90 0 2 0 3 9 0 22 tive ratio, in contrast to the teachers, whogenerally balanced out the positive and negative statements. Evidently it waseasier for students to give a negative verdict. In humorous statements,the boys gave 233 while the girls gave 96 in the same period of time despitealmost equal representation in the class. Similar differences occurin each content category favoring the boys at about a 2 to1 ratio. Among the results for boys and girls on cognitive memory,the ratio brAween clarifying meaning and clarifyingqualification was almost even in all classes. Yet, inteacher stateatents, as shown in Table 9, the ratio was at least 2 to 1, sometimes much greater, in favor ofclarifying meaning. Apparently, students engaged more incorrecting or adding to original statements than inreformulating or clarifying ideas. Along with the high number of negative verdict responses,this fact suggests that the students were spending considerable time inclass disagreeing with, or amending, other students' and teachers' statements.Observers suggested that some gifted students take this relatively easy way out-- making minor corrections or amendments as their contribution toclass discussion rather than generating ideas or concepts of their own,which require preparation and cognitive thinking. Convergent thinking responses of the boys and the girlsfollowed expected trends. The ratio of production was 2 to 1in favor of the boystypical of the overall results. The different styles of the twosci- ence teachers were noted toaffect the performance of the students. In the FOX-GEORGE groups there were aboutfive times as many responses as in DAN-EASY. Teacher statementsand questions in Table 9 showed the DAN-EASY teacher making more statementsbut phras- ing fewer questions in this thought dimension. This wasanother indi- cator of how closely student response wastied to teacher behavior. In the area of divergent thinking, the totalperformance of the boys seemed to outweigh that of the girls in all sections exceptJACK-KING, in which the teacher's disinclination to encouragecognitive behavior apparently limited the performance of both groups. Most ofthese same students were also members of the FOX-GEORGE groupsand participated in divergent thinking operations in thesesections. The category of synthesis was not often found. It appearedthat ex- pecting such high level abstracting behavior in oral discussion maybe unrealistic. Gifted students perform such intellectual feats wellin written work where time, prior planning, andclass discussion have laid the groundwork. Results of implication type questions show a range of boys' responses extending all the way from 0 in JACK to 67 in CHARLIE-Fall.Girls showed a similar, if more restrained, response. Thus, relativelyfew

52 the part of the stu- requests in this areagenerated much reaction on dents. be- Responses in evaluativethinking showed arather equal division ratings and structuredprobability statements. tween unstructured and some zero scores were Fewer comments weremade on qualification since one can't quarrelwith an unexpressed obtained in this category few rat- rating or judgment;for examples inDAN-EASY sections, very thus few counterjudgments expressed.There ings were made and think- seemed to be moredifferences among classsections in evaluative this was broughtabout by one ing than in othercategories. Generally, the stu- student making ajudgmental responseand the rest of dents chiming inwith judgments of their own. the performanceof the studentsseemed highly In summary, while remained related to teacherperformance over allcognitive areas, there the same teacher.This impressive differencesbetween sections with there were morecomplex factors,involving the suggested that perhaps the par- composition of thegroup and perhaps eventhe accident of statements ticular topic ofdiscussion, influencingthe proportion of made in each category. obvious variations amongthe teachers, all ofwhom In view of the empirical had been rated aseffective teachers bytheir supervisors, effectiveness seems desirable as a study of just what is meantby teacher example, with oneparticular style such asteaching first step. For probably be necessary phonics or using thediscovery method, it would depth the actualperformance of individualteachers, to analyze in some would be similar be- rather than to assumethat classroom behavior the same back- cause theyhad the samephilosophy of teaching or ground of training.

Relationship between Indusand Extraclass Factors Tables 11 and 12 showthe results of the principleaxis component anal- rotation) computed todetermine the relationshipsbe- yses (varimax variety of ex- tween scores onthe classroominteraction system and a ternal variables. Thesample was divided intoGroup A and Group B intrusive influence. Inaddi- to keepchronological age from being an data were analyzedseparately. The component tion, boys' and girls' The analysis is very similar tothe more commonlyused factor analysis. difference lies in the valueinserted in the diagonalof the matrix. In it is the sums of the squares.The reader is in- the component analysis that one would vited to interpret componentsin the same framework interpret the more commonlyused term, factor. Group A Results. Table 11shows the componentloadings obtained

53 _ ; = U1 Group A Principal Axis Component Loadings by Sex (Varimax Rotation) Bo s - TABLE 11 Girls ClassroomVariables Performance DivergentCognitive Memory 1 7776 2 3 34 4 5 36 6 7 85831 2 3 4 5 6 7 Test Data VerbalEvaluativeConvergent IQ 8488 79 8881 86 37 ConsequencesUsesNonverbalConsequences Breadth IQ Breadth S/T 77 9353 87 AP 7 3035 84 8161 87 Semantic Differential FatherMotherStudent 798483 527082 38 31.39 31 CompetitionWorkSuccess 317552 32 41 33 726765 35 43 a. SchoolImaginationLoveFaith 653244 33 37 60 37 62 78 46 655256 30 4385 39 Teacher Rating CognitionSociability 56. 42 76 31 38 .72 31 3C PercentChronological Variance Age 27 2231 12 4711 10 SI10 tI 24 24 70 16 .8350 11 10 8 E TABLE 12 Variables Group B Principal Axis Component 1 2 3 Bo s 4 5 Loadings by Sex (Varimax6 Rotation) 7 1 2 3 Girls4 5 6 7 Classroom Performance EvaluativeConvergentDivergentCognitive Memory gi909292 948183 Test Data Verbal IQ _ 32 68 40 52 ConsequencesUsesNonverbalConsequences Flexibility IQ Flexibility S/T 6153 34 76 4585 37 36 8742 57 49 6730 38 9C Sentence Completion WorkFearsTimeSelf Styles 64 79 4377 90 5586 82 70 30 Teacher Rating IntellectualOtherFamily People 55 7938 35 33 8681 40 60 tll PercentChronological Variance Age Sociability 27 72 15 13 12 12 49 12 89 9 23 19 14 12 8948 11 1075 10 *1111.1

in the Group A analysis.This group was composedof students in BAKER-CHARLIE and HAT-IDEAsections. In both Tables11 and 12 only component loadingsof .30 or greater werenoted. Component I loadings of the girls andComponent II of the boyscontained the major for all classroom variables. The remarkably highloading of all classroomvariables on one com- there ponent came assomething of a surprise. It hadbeen supposed that per- might be two or more componentsabstracted from the classroom formance data, with stuck ntsrevealing different cognitivestyles and preferences. However, the common loadings canbe explained by thevariety of intellectual demands made inthe classroom environment.Suppose the divergent thinking question,"What might havehappened if the Spanish Armada had won overthe British?" were asked.Any student answer to thisdivergent question mustnecessarily involve convergent thinking and cognitive memoryskills as well. The student must re- member what he has previouslylearned and then reasonfrom the proposition to a conclusion. Inaddition, evaluativethinking would be induded in the student's choice ofwhich answers he deemedappropriate. In other words, the studentwho wishes to expresshimself in the classroom is forced to useall the major categories ofthinking operations. Once this fact is recognized,it is not unreasonable to expecthigh intercorrelations among the classroomvariables. This is what wasobtained. In the Group A boys(and to a lesser degree theGroup A girls) classroom variables werefound related to the teacher'sratings of cognitive ability. A number of othervariables showed small loadingswith the classroom variables in the GroupA girls sample. Theseincluded a negative loading for the conceptsof Faith and Student and apositive loading for Competition. Thedivergent thinking scores onthe Uses and Consequences tests load at alow level on this generalexpressiveness component for the girls. Component I for the boysand Component II for thegirls loaded heavily on the semantic differentialconcepts. Of the ten conceptsin- volved in the semantic differential,only Faith in the boys andImagina- tion in the girls were absent fromthis component. It may alsobe noted that in neither the boys' Orgirls' case does this general component, which might be labeledsanguinity, have any relationship toother variables. With the boys, Component III washeavily loaded with verbal and nonverbal IQ scores, with a smallpositive weighting on teacherrating of cognition, as might be expected,and a small loading on thesemantic differential concept of Love,which lacks an obviousinterpretation. Component IV for the Group Aboys can be labeled thedivergent

$6 and Uses breadth thinking factor, loadingheavily on the Consequences scores as well aschronological age. Divergentthinking performance showed a small loading.While this was anexpected in the classroom also small result, it was the only oneof the four groups inwhich even such a relationship was found. Component V for the boysrevealed the Consequencessolution score of Love on the loading highly, with anegative loading on the concept high percentage semantic differential. Inother words, boys who gave a positive of solution answers tothe Consequencesquestions were less combination to explain. toward the concept ofLove. This is a difficult Component VI for the boysloaded most heavily onthe teachers' rat- this was performance ings of sociability of thestudent. Connected with Uses breadth, and on the cognitive memorydimension in the classroom, of Imagination. It is tobe noted that a positiveloading for the concept the ratings oncognition and sociabilitywith the the teacher separated with the boys, but that theseratings are found onthe same component girls. Component VII for GroupA boys loaded mosthighly on Faith, with antagonistic to negative loadings forImagination; these concepts are The higher statusheld for the conceptof Faith and the one another. repeated in lower status of Imagination, aninteresting finding, was not the girls' sample. Component III for the GroupA girls, accountingfor 16 percent of the variance, was the IQfactor and was associatedwith a strong negative loading on chronological age.In other words,the younger girls appeared to have a higher IQ.Also, small negativeloadings were found to mean thatthe for the concepts of Schooland Work. This appeared higher IQ girls did not rateSchool and Work ashighly as the other attitudes girls, who may beconsidered as havingremarkably positive on these two concepts. Component IV for the GroupA girls representedteacher attitude the older girls more and ratings. Here again,the teacher tended to rate positively. Only minor loadings onthe semanticdifferentials of Stu- teachers saw the girls' dent and Faith accompaniedthe ratings. The cognitive and sociabilityratings as closelyintertwined. Component V for the girlsobtained its greatestloadings on the se- loading was also ac- mantic differential conceptof Imagination. Such companied by a positiveattitude towards the conceptsof Love, Work, Student, and Father, and asmall loading on convergentthinking performance in the classroom. W 0.hasten to point outthat having a high having a pro- regard for the concept ofImagination is not the same as ductive imagination, andthat this component seemsmoreconservative and convergent in totalcharacter. 57 Component VI for the girls was the divergent thinking factor corre- sponding to Component IV in the boys. Only negative loadings towards the concepts of Love and School may be noted here. High divergent thinking would be expected to show negative loadings to traditional value concepts such as these. Finally, Component VII for the girls was represented by the solution scores on the Consequences test which did not load substantially or meaningfully with any of the other variables. In summary, the two components with the largest loadings for both sexes represented classroom performance and semantic differential scores. There are few indications that performance on divergent thinking tests is related to such performance in the classroom. IQ and divergent thinking scores fell in separate factors and neither seemed highly related to the attitudes toward semantic differential concepts. Compo- nents IV to VII in both groups accounted for about the same amount of variance with rational interpretation becoming more difficult as the variance contribution became less. Group B Findings. Table 12 shows the results of the principal axis component loading from the sample of students in the DAN-EASY, FOX-GEORGE, and JACK-KING groups. In this sample, a sentence completion test was used to obtain scores on self concept, attitudes toward family, other people, time, work styles, and fears. Component I for both boys and girls in this group was classroom expressiveness. As in the Group A findings, all of the classroom variables loaded heavilyon the same component for both sexes. The teacher's rating of cognitive ability in girls loaded on this component but no variable loaded for the boys on Component I. Component II for the boys accounted for 15per- cent of the variance and appeared to be loaded with divergent thinking, highly related to the sociability variable. The teachers rated boys scoring high on divergent thinking tests as most popularnot as the rebel that he has been pictured in other research. Also included in this component was a more negative attitude towards other people and such concepts as imagination and creativity from the sentence completion test. Component III for the Group B boys carried the highest loadings on self concept. Also associated were less positive attitudes toward future or past events and creativityimaginative work styles. A positive self image in these boys did not mean positive attitudes toward the other concepts. Component IV for Group B boys loaded on Consequences solution score and also related to positive attitudes towards family in the sentence completion test and small loadings on. verbal IQ and teacher ratings of cognitive ability. Thus, ability to produce solutions to consequences

58 . <4,,,,,,A7r,

questions was related to a generallypositive attitude towards closefam- ily and was reflected in other persons'judgments of the boys. Component V for the boys was the IQfactor which included a small loading from the divergent thinking tests.This was the only time there were positiverelationships of a significant naturebetween IQ and divergent thinking tests. Component VI for the Group Bboys indicated a relationshipbe- tween absence of fearsand positive ratings of intellectualperformance by the teacher, a finding whichmight have been expected. Component VII appears to be mainly achronological age component. With the Group B girls ComponentII seemed to represent thegeneral sanguinity factor found in the GroupB boys and girls. However, none of the other variables seem torelate to it. On Component III forthe girls the major loadings are forthe divergent thinking testwith some small loadings on verbal IQ and apositive loading on teacher ratings of cognitive ability. This seems toindicate that girls who show positive performance on divergent thinking testsalso do well on IQ tests and are perceivedby the teachers as high achievers.Another unusual aspect of this component was thenegative loading of nonverbal IQ scores. On Component IV, positiveattitudes toward self appeared,along with an absence of fears.However, the higher the conceptof self, the lower the flexibility score onthe Uses test, a relationship similar to that found with the boys ofthis particular group. Does this meanthat a certain amountof anxiety and a limited self concept are necessaryfor extensive production of ideasin divergent thinking? On Component V, the majorloadings were in the area of teacher ratings. The teacher ratings werepositively related to nonverbal IQ scores of the youngsters and aslightly positive self concept on the partof the girls. As with the Group Agirls, the teachers seemed to seecognitive ability and sociability as being moreclosely related in girls than in boys. Component VI was the IQ factorrelated to chronological age and, in this case, a negative loadingbetween performance on the IQ test and Uses flexibility. None of theseweightings are very strong on the Uses flexibility. It is likely that the strongCA loading had an influence on this component. Component VII, the last factor forthe Group B girls, was the Con- sequences solution scorewhich seems related to performance on nonverbal IQ tests, with no relationship tofamily attitudes, in contrast to Component IV for the boys. The failure of the componentanalyses to yield comparable components in the two groupsshould signal caution to any attempt to gen- eralize from the present findings. The two groupsdid have in common a single factor ofclassroom expressiveness including allmeasured var-

59 iables of classroom performance, which in turn, was notrelated in great degree to other outsidevariables. Divergent thinking was found to be a component by itself,separated in most cases from IQ scores. There was some indication in Group B thatdivergent thinking was related more to affective variables than tocognitive variables, a point that will be dealt with in the discussion.There was evidence of different patterns of relationshipsbetween the sexes once the first two components were eliminated.

Relationship between Cognitive Performance ofTeacher and Student The percentages of teacher questions ineach of the primary categories were ranked from one to ten.In one class, BAKER-CHARLIE, where the teacher was recorded in 20 sessions,the ranking was from 1 to 20. The same class sessions were rankedby sex in percentage of production in each of the thought categories.For instance, the ten sessions of FOX-GEORGE were ranked in percentage ofdivergent thinking questions presented by the teacher; theseclasses were then ranked in percentage of divergentthinking presented by the boys and girls. The relationships between setsof ranks obtained for each teacher's class sections were measured by aSpearman rank order correlation co- efficient. Table 13 indicates aconsistently high relationship between

TABLE 13 Spearman Rank Order Correlations of StudentPerformance versus Teacher Performance Number of Cognitive Divergent Convergent Evaluative Class Memory Thinking Thinking Thinking Sessions*

.75 .78 .65 .50 Boys 20 Teacher A vs.Girls .58 .72 .48 .47

.57 .74 .76 .80 Boys 10 Teacher B vs.Girls .57 .08 .43 .38 .69 .51 .71 .68 Boys 10 Teacher C vs.Girls .90 .71 .91 .55

.77 .80 .77 .62 Boys 10 Teacher D vs.Girls .77 .62 .48 .43 .39 .82 .79 Boys .82 10 Teacher E vs.Girls .88 .60 .83 .56

* For N = 10, Spearman r's of .56 significant at .05level (cae tailed test) N = 20, Spearman r's of .38 significant at .05 level (one tailed test)

60 -p ,e7

kind given by the the thought processes askedfor by the teacher and the student. In most instances,the positive relationshipreached a level of statistical significance. In noinstance was a negativerelationship found. In only one case did therelationship approach zero,when teacher B's request for divergentthinking was related tothe divergent thinking production of the girls. Acloser examination of theseparticular class sessions showed that theboys responded mostconsistently to requests for divergent thinking asshown by an r of .74,whereas the girls appeared somewhat erratic intheir responses. Withthat single exception, no substantial differences appearedbetween boys and girlsin their respon- siveness to the type ofquestion asked. In retrospect,this was not surprising. If a teacherasks for the date Lincoln wasshot, it would would ue difficult for astudent to give adivergent thinking response. If ateacher asks for comparativejudgment on the communication effectiveness of thenovel versus that of theshort story it would be hard to avoid anevaluative response. In short,the teacher has dominant classroom. Such a fact control over the type ofexpressive thought in the presents teacherswith special opportunitiesand special responsibilities.

Children over Change of Consistency of ClassroomExpressiveness in Gifted Time, Subject Matter, andTeacher Table 14 gives an indicationof the consistency ofclassroom expres- different times in siveness. A number of thestudents were recorded at and teacher. the same class, or in differentclasses with different content performance and In view of the observedrelationship between teacher student production, there was somequestion whether interstudent con-

TABLE 14

Consistency of ClassroomExpressiveness in Gifted Students Social Studies (Fall)Social Studies Science vs. vs. vs. Science English Cognitive Social Studies (Spring) (N = 20) (N = 39) Variable (N = 39)* .66 .60 Cognitive Memory .40 .44 Divergent Thinking .31 .31 .46 Convergent Thinking .37 .58 .45 .52 Evaluative .54 * For N = 39, Pearson r of .30significant at .05 level N = 20, Pearson r of .43 significantat .05 level (Guilford,I950b, p. G09)

61 sistency could be noted from one period to another, orwhether student performance depended entirely on the teacher's style. Data provided by three samples of children wereanalyzed. The first group contained 39 students from twosocial studies classes. Two sets of observations were taken with the same teacher; oneblock of recordings was made in the fall and theother in the spring. A second group of 20 students was observed during social studies andagain in science class. A third group of 39 students, originally recordedin science classes, were recorded a year later in English classes. Pearson product moment correlations werecalculated between adjusted class scores in each of the groups.Table 14 shows results of these correlations. One striking fact noted was thepositive nature of the correlations. Regardless of the class or thecognitive variable, there was significant student consistency in terms ofclassroom expressiveness. In other words, despite the clear influence ofthe teacher on the thought processes, the studentsstill maintained, to some extent, their own consistent styles. As shown by the science versusEnglish comparison, students were consistent even while moving from oneclass to another, more so than students remainingin the same classes with the same teacher. Of the four major categories, divergentthinking appeared to have the least stability. This was consistent with theobservation that it was the most susceptible of the primary categories toteacher initiation and influence. Cognitive memory, which seemed to be a necessary partof cognitive interchange in any classroomsession no matter which teacher or subject was involved, maintained thehighest level of student consistency from one time to another.

Daily Consistency of Teacher Student Expressiveness within Class Sections While Table 14 indicated student consistency ingeneral expressiveness, there was a further questionregarding the consistency of the level of expression of various thought processes within aparticular class section. In order to calculate whether teachers orstudents varied from expected level of performance during a five dayweek, the following procedure was used: a percentage was calculatedfor each of the thought processes used for each class session. A chi square wasobtained using for expected frequency the total percentage forthe week divided by five. The null hypothesis stated that if nodifferences occurred in the proportions of thought processes from one classsession to the next, there would be relativelylittle difference between expected and obtained frequencies for each of the class sessions. Thenull hypothesis would be rejected if large differences were foundbetween obtained and expected performance.

62 Differences by Chi Square on Proportions of Teacher Cognitive Expression within Five Teacher Questions TABLE 15 Teacher Statements Class Sessions Groups CognitiveMemory DivergentThinking ConvergentThinking EvaluativeThinking CognitiveMemory DivergentThinking ConvergentThinking EvaluativeThinking 5.00 CHARLIE-FALLBAKER-SPRINGBAKER-FALL 1196**14.96**6.261.65 41.80**18.37"12.80*4.43 22.71**14.21**11.31*1.08 23.20* 4.508.444.60 5.661.933.394.11 4.505.174.171.20 10.58*2.386.703.43 7.575.503.46 EASYDANCHARLIE-SPRING 6.131.55 17.20** 2.50 19.55**6.78 7.609.33 6.471.161.25 9.004.33 .50 24.91** 2.09 .83 19.25**2.626.17 GEORGEFOXIDEAHAT 34.55*23.35*10.24* 9.52* 37.33**13.00*16.50**6.33 22.1324.12*14.344.36 31.47**21.00**16.83**4.33 5.671.52 .93 6.506.751.67 20.50*10.39*4.35 3.502.186.14 JACKKING Significant at .111.05 level level (x2 (x2 = = 9.49) 13.28) 31.21* 7.14 29.00** .00 57.15" 7.44 45.37**13.00* 5.312.10 2.003.50 4.291.87 8.445.92 03Of 1,1011 Table 15 shows chi square findings within the five classsessions (see also Figures 3-8). The results indicate clearly that,regarding types of questions asked by teachers, proportions varied considerablywithin a particular class. They did not necessarily vary in the same wayfor each class section. For example, for the teacher in theBAKER section, the divergent thinking questions varied significantlyin the BAKER-Fall section, while the cognitive memory and convergentthinking questions varied significantly for the BAKER-Spring section. Significant differences were obtained on divergent thinkingfor the DAN section and on convergent thinking for theEASY group. It should be recognized that these results were not independentof one another since they were composed of proportions thatadded up to 100 percent. A high. percentage of divergentquestions on a given day in- evitably meant that proportions would be lower in the other areas. The general portrait was clear. Teachers varied questionasking from one time to another within the sameclass, depending upon the goals for a particular session. When teacher statements were considered, a greater degree of consistency was obtained. In the BAKER-CHARLIE series, only convergent thinking statements made in the CHARLIE-Springsection reached a level of statistical significance. Onlyin the FOX-GEORGE sections was there some degree of consistency from oneclass to another. In both sections, there was significant variation in the proportion of convergent thinking statements made in the five sessions. This was probably because teacher explanations necessarily accompanied the science experiments performed only on certain days. In the JACK and KING sections, the teacher remained consistentin kinds of statements made over the entire five days. This suggests that the teacher's variations in expressive thought relate much more to questions than to statements. Different patterns of classroom operation were seen as relating to kinds ofquestions asked. Variation for students, both boys and girls, in the five sessions is shown in Table 16. Chi square was used in the same way as with the teacher data. Question asking and statement making data were combined for reasons previously stated. Wide variation again appeared and, for the boys, especially in divergent thinking. There were significant ;)- intersession differences in the BAKER-CHARLIE series, and in one section of each class except for JACK-K1NG where, it may be recalled, very little divergent thinking was expressed at anytime by the boys. In general, the student variations were rather closely tied to variations in teacher questions. Inspection of the girls' performance suggests that they varied even more from one session to another than did theboys, with a greater num-

64 - Differences by Chi Square on Proportions of Student Cognifive Expression TABLE 16 Boys' Statements Convergent within Five Class Sessions Evaluative Cognitive Divergent Girls' Statements Convergent Evaluative BAKER-FALLBAKER-SPRINGGroups CognitiveMemory 28.10**10.48* DivergentThinking 32.8210.43 Thinking 12.93*6.26 Thinking 22.8710.46* Memory 33.9534.54 Thinking 57.9525.37 Thinking 27.0010.54 Thinking 25.0024.50 EASYDANCHARLIE-SPRINGCHARLIE-FALL 16.84**2.187.591.71 25.0927.86"16.13**8.20 8.898.924.353.00 14.009.405.757.20 29.31 6.633.302.69 33.3611.67*14.004.00 32.4812.33*11.36*9.86 12.80*10.1010.40*11.08 FOXGEORGE 5.985.02 12.29*8.798.40 20.22**8.148.41 2255.60 4.658.27 13.009.00 21.8719.44 4.006.40 KINGJACKIDEAHAT 26.57**24.9614.56" 5.07 11.09*3.002.50 38.27**19.11" 9.80* 21.78**30.27" 2.739.43 24.6224.3036.33 .66 11.00*14.0053.1711.00 33.2028.0012.296.62 22.6742.0832.006.35 CD ** Significant at .01 levelSignificant at .05 level (x2 = 9.49) (x2 = 13.28)

, ATP',

ber of significant differences obtained across all groups. Again, the relatively few divergent thinking statements made by girls sharpened the differences between expected and obtained performance. These results indicate that a sample of teacher behavior needs to be relatively comprehensive since it cannot be assumed consistent from one class session to the next, particularly in regard to questions. (The same does not seem true of teacher statements, for the pattern remained relatively consistent except for special circumstances.) Thus, teacher performance must necessarily be observed over a relatively long period of time. The students' cognitive performance reinforces this statement even more strongly. The student performance, in propot don of thought processes, varied by the day and was affected substantially by changing circumstances and class goals. In effect, these results represented statistical confirmation of the variance that was observed.

Factors Influencing Type of Production in Class Table 17 shows the results of a three way analysis of variance investigating whether teacher performance in the primary thought categories was related to the group, the class section, or day of the week on which the recording took place. A group refers to all students assigned to one teacher, such as FOX-GEORGE or JACK-KING. With five teachers and six groups involved, the degrees of freedom for groups was five. The investigators wished to determine whether significant variation occurred in various areas from one class session to another. Since suggestions had been made that teacher and student styles of performance might be influenced by day of the week, this was also included as part of the total analysis of variance. Table 17 shows the results of the analysis on teacher questions. Signi- ficant differences were obtained on all primary thought categories on the basis of group. This was confirmation of what many observers have indicated, that teachers vary significantly in style of question asking. For the area of cognitive memory, the level of statistical significance was reached when class sections were compared. As noted in Figure 7, there were particularly dramatic differences in the two social studies sections, HAT and IDEA, which may well account for the significant result by class section. In the productive thinking categories, no significant variations were found due to class section, day of the week, or the interaction among the basic variables. Of the three variables, the difference from one group to another was by far the most significant in accounting for variation in the primary thought categories. Table 17 also shows analysis of variance using teacher statements as the basis for comparison. In the same way, variations due to group,

66 Analyses of Variance of Thought Processes Expressed by TABLE 17 Source df Mean Square Cognitive Memmy F Mean Square Divergent ThinkingGroup, Class, and Day Teacher Questions F Mean Square Convergent Thinking F Mean Square Evaluative Thinking F Group-ClassDayClassGroup 451 936.15216.31159.81666.42 7.65*5A4**1.771.30 99.1811.02 2.37 .00 3.64* .10.00.40 1391.15448.271¢6.7246.31 14.58**4.701.74 .48 201.66 27.8198.8243.50 5.932.901.281.81 TotalResidualClass-DayGroup-Day 5920 4 122.42107.36261.91 2.14 .87 27.1929.8929.13 Teacher Statements 1.101.06 191.56 95.4131.72 2.01 .33 34.0268.6929.50 2.02 .87 GroupGroup-ClassDayClass 5415 132.0253334 76.8285.40 17.98"2.592.884.45 37.31 3.134.271.44 14.51" 1221.66 .56 64.1238.4094.3116.92 2.531.523.72* .67 108.51 54.5115.75 7.35 7.81*3.92*1.13 .53 TotalResidualClass-DayGroup-Day =p

dass section, and day of the week were considered together with all of the interactions of these variables. The results were somewhat similar to those found in analyzing teacher questions. Significant variation was found among groups on all of the primary thinking processes, butno significant variance was found between class sections for any of the thought operations. Neither were there significant variations due to the day of the week. Two significant interactions were obtained. One related to the combination of class session with the day of the week in divergent thinking, and another, in evaluative thinking, to combining group with classsessions. It was felt that the small number of certain types of statements by some teachers made this significant interaction possible by unduly weighting the day of the week when this series of responses occurred. No clear explanation can be given for the interaction betweengroup and class on evaluative thinking except the suggestion that the makeup of the youngsters in the class molds the behavior of the teacher and forces the teacher to modify style to allow for student idiosyncracies. Table 18 indicates the analysis of variance on the boys' statements for the four major areas of thought processes bygroup, class section, and day. Significant differences were found in theareas of divergent thinking and convergent thinking by group, again underlining the differences among groups. Two significant differenceswere found due to the class section in the areas of cognitive memory and evaluative thinking. It will be recalled that in analysis of teachers'statements and questions (Table 17) there were significant differences by class section in cognitive memory questions and in evaluative thinkingstate- ments. The finding of potential influence of the teacher upon the kinds of statements made by the students is thus supported. Again, neither the day of the week nor the interactions revealed significant variance influences. Table 18 also shows the analysis of variance obtainedon thought processes of the girls. Results were consistent with the responses of the teachers and the boys in that the main significant variancewas obtained between groups. In the girls' case, neither the sectionnor day of the week had significant influence, nor were significant interactions found among the major variables. In total, the major variance in expression of kinds of thoughtprocesses related most strongly to the specific group, which in this case meant the particular teacher. Occasionally, the membership of a class section appeared to influence the kind of expressive thoughts of teacher and boys, but not of girls. Day of the week was not significant nor was variable interaction particularly impressive for student or

68 TABLE 18 Cognitive Memory Analyses of Variance of Thought Processes Expressed by Divergent Thinking Group, Class, and Day Boys' Statements Convergent Thinking Evaluative Thinking GroupSource df Mean15 Square 319.15 2.90 F Mean Square 530.42 12.52* * Mean Square 931.39 14.76** F Mean Square 109.71 5.78 F Group-DayGroup-ClassDayClass 20 54 1306.67226.15134.22 96.56 11.87**2.051.22 .88 74.7827.2240.02 3.43 1.76 .64.10.94 1M.84340.8284.7867.90 2.121.341.085.40 123.2726.9339.5450.19 2.081.422.646.50* TotalResidualClass-Day 5920 4 110.04 61.08 .55 49.3536.60 Sia:ements .86 63.1176.15 1.20 52.8918.97 2.79 Group-ClassDayClassGroup 5451 1083.56157.02169.9726.67 1.087.46**1.17 .18 627.55 30.4717.07 7.51 8.71**.10.42.24 133.96890.8355.39 2.40 1.218.06** .50.02 188.1250.4737.81 .27 1.043.91* .78.00 TotalResidualClass-DayGroup-Day 205920 4 257.21309.14145.17 2.131.77 72.0375.4485.51 1.051.19 110.55167.44139.40 1.511.26 48J22.5636.33 6 .46.75 ** * = p < p < .01 .05 I; teacher questions, although some interaction was found in the teacher statements in divergent thinking and evaluativethinking.

Sex as Relevant Variable in Classroom Expressiveness and Other Comparisons Table 19 shows results of comparing boys and girls in amount of cognitive expressiveness related to the various thought categories in the six sets of recordings. The mean scores in Table 19 represent the average number of responses made in each of the four majorcategories during the five consecutive sessions. A casual perusal of the table shows that generally the boys were more expressive than the girls over all categories. However, such a statement needs qualification. The BAKER-CHARLIE Spring group had nearly the same members as the BAKER-CHARLIE Fall classes. The spring data is really an indication of the stability of the fall observations and should not be considered data on a new group. In the BAKER-CHAR- LIE Fall group, clear differences between boys and girls were found in favor of boys' expressiveness in all areas except evaluative thinking. However, the boys appeared superior in the evaluative category in the spring sample. Some of these same youngsters (BAKER-CHARLIE) were in the DAN-EASY group and may well have been responsible for some of the significant differences shown in Table 19 for the DAN-EASY group. The FOX-GEORGE group, however, was a completely new set of students. The data indicate that the boys were significantly more expressive than the girls in every area. Even in the categories of evaluative thinking and divergent thinking, infrequently called for by the FOX-GEORGE teacher, the boys dominated the few statements made. The HAT-IDEA social studies classes, theoretically similar to the BAKER-CHARLIE groups, did not show the same superiority of boys in expressiveness, despite the fact that these students were at the same age level as the BAKER-CHARLIE group and presumablyfrom the same general population. Only in the area of divergent thinking were differences found in favor of the boys, While this supported the general notion that boys were more inclined to divergent thinking, the difference in total expressiveness was negligible. In the JACK-KING group, there were significant differences in favor of the boys' expressiveness in the areas of cognitive memory, evaluative thinking, and in total responses. There was also a tendency for boys to be more expressive than girls in divergent and convergent areas, but these differences did not reach expected levels of statistical significance. Within the limits stated above, there was sufficient evidence to state that boys tend to be more expressive in the classroom in all of the pri-

70 TABLE 19 Groups Mean CognitiveMemory Verbal Classroom Expressiveness inS D Gifted Groups by Sex Mean DivergentThinking S D ean ConptergentThinking S D Mean EvaluativeThinking S D Mean Total S p BAKER-CHARLIEBAKER-CHARLIE Spring FallGirlsBoys (N=20) (N=23) 16.78"9.50 3.959.94 5.309.00" 3.895.74 4.807.43" 3.094.91 3.153.35 2.603_25 22.7536.56" 10.9719.15 DAN-EASY BoysGirls (N=17) (N=14)(N=22) 27.71"11.5926.27" 21.451632 9.17 4.35"4.127.86" 4.944.005.60 11.14"6.068.71 4.755.735.95 2.235.41"1.71* 2.133.13424 26.0049.77"40.57" 27.9326.2018.39 FOX-GEORGE BoysGirls (N=19) (N=28) 25.36"10.8212.84 27.3015.7611.03 1.003.25" .81 2.003.58 .99 21.14"6.642.58 11.2222.30 2.93 225" .5438 2.431.32 .69 19.0052.00"16.95 29-2553.6714.08 HAT-IDEA BoysGirls (N=16)(N=28) (N=24) 17.4416.04 10.8112.67 2.814.88* 2.205.07 5.696.33 5.636.36 3.633.42 3.593.02 295630.66 20.162421 JACp K-K< .01 INGop < .05GirlsBoys (N=25) (N=23) 14.52*7.52 15.04 9.92 .32.52 .63.73 10.096.60 7.789.05 1.04226* 2381.67 2632*15.48 235718.84 mary categories than do girls. In no single instance in this study did the girls even approach statistical significance inmanifesting greater expressiveness than the boys inany category. However, since factors relating to classroom expressivenessare obviously many and varied, it should not be expected thatevery class or group would show the same results. Since there was considerable overlappingof membership among groups, a further step was taken to determine if therewere true sex differences in classroom performanceover the entire sample. In order to equate performance in various classrooms, since therewas a wide di- versity of production fromone class to another, a weighted score for each student was determined. Each student's totalproduction for the five day period was divided by the totaloutput of his class and then multiplied by one thousandto eliminate decimals. An index score of the proportion of the cognitivememory responses, divergent thinking responses, etc., given by each student was obtained. Whena student appeared in more than onegroup, results of performance in the first group were used. The results of this analysis (Table 20), comparing 79 girlsand 86 boys, were quite unequivocal. On all primary thoughtcategories, the boys were found to be more expressive in the classroom thanthe girls. On evaluative thinking the difference obtainedwas beyond the .05 level of significance, but beyond the .01 level for the other three areas. As far as the present sample was concerned, it can be stated that gifted boys were more expressive in all thought dimensions than the girls in the classroom discussion environment. Table 21 shows sex differences in the current populationon test measures and teacher ratings. Results are divided into Groups A and B on the basis of age and different measures in the attitudinalarea.

TABLE 20

Comparison of Boys' and Girls' Adjusted Scores on Classroom Perforniance Girls (N=79) Boys (N=86) Mean SD Mean SD Cognitive Memory 27.24 R5.60 42.60 34.64 3.23 <.01 Convergent Thinking 26.48 29.33 47.16 41.32 3.71 <.01 Divergent Thinking 26.02 32.26 52.56 52.16 3.94 <.01 Evaluative Thinking 31.16 3739 43.90 42.74 2.02 <.05

' Sex Differences on Test Measures and Group A TABLE 21 Teacher Ratings Group B CA in Months Measures 130.56147.33MeanGirls (N = 36) 10.17 4.88SD 13153147.26MeanBoys (N = 47) 9514.24SD 163.09129.38MeanGirls (N = 56) 10.15 7.09SD 163.62129.76BoysMean (N = 50) 8.18724SD VerbalConsequencesUsesNonverbal IQBreadth IQ Breadth 125.75 13.2815.9243.28t 112219.253.003.67 129.68 35.7914.0017.19 18.873.763.739.43 129.1821.0445.89°13.80 11.8913.88 3.167.36 131.8622.5639.7813.64 10.6316.40 4.096.58 Consequences Solutions FathersMothersStudents Differential)(Semantic 70.44t71.83t72.7867.06 8.835.089.15858 672166.9467.747153 10.03 9.276.82 WorkSuccessCompetitionLove 71.6771.6167.9267.75 9.308.627.617.44 65.9871.3470.1364.15 12.4010.9511.91 7.78820 SchoolImaginationFaithSelf (Sentence 69.4771.42* 8.17658 69.1966.81 105712.02 23.32° 4.05 3.82127 20.08 3.92 8.82123 FamilyWorkTimeFearsOther Style People Completion) 13.32°12.95t12.98 5.89 1.813.433.402.96 14.1014.5212.386.16 1.721.822.293.01 tai t P < SociabilityCognition *P< Rating)(Teacher p < 11.42 2.22 3.86 .68 11.87 2.34 3.69 .89 11.07t 1.84" 4.33 .65 9.68224 3.63 .80 As reported in the earlier tables, there were no meaningfuldifferences between the sexes in verbal or nonverbal IQ, with the average score on group intelligence test measures well intothe expected range for a gifted population. The results on divergent thinking were of particular interest since this appeared to be one of the more decisive areas of difference. The question was whether this observed classroom difference was due to differences in basicthinking abilities of boys and girls, or rested in the greater willingness of the boys to expressideas, with a greater role expectation for aggressive performance in class.If basic differences were in thinking ability, the boys should be superior both in classroom expression and in paper and pencil tests of divergent thinking; but if role expectation were the determining factor, no sharp differences should be found between the sexes on thewritten tests. The evidence suggests that the latter explanation has some merit. There were no significant boy versus girl differences between the mean breadth scores on the Uses and Consequences tests in either Groap A or B. However, differences approaching significance were obtained for both groups on the Consequences solutions scores. In both groups,the girls contributed a significantly greater proportion of solutions than did the boys. Solutions represent a more intellectually mature response, and in this area, the girls seemed superior. On the measures of attitude and self concept, the results were fairly consistent with previous studies on sex differences. In group A, with the use of the semantic differential, the girls appeared more positive toward most concepts than did the boys. Since the highest positive score obtainable was 84 and the midpoint score was about 50, it can be seen that all groups scored in a strong positive direction. However, Group A girls were significantly more positive at a probability level of .10 on the concept of Fathers and at the .05 level on the concept of Imagination. The boys rated the concepts of Competition and Success less positively than concepts of Faith and Mother. No significant difference was found, however, between the boys and the girls of Group A in the teacher ratings of cognition or sociability. In Group B, on the sentence completion measures, significant differences were obtained between boys and girls in three of six general areas. Since low scores represented a positive orientation, the girls appeared to have significantly lower self concepts, feeling less positive about themselves than did the boys. On the other hand, the girls were more positive towards family and other people. No differences were found in number of fears, time orientation (positive attitude to future or past), or attitude toward concepts such as Creativity and Competition. In a sense this confirmed some of the semantic differential results obtained for Group A, suggesting that the girls were slightly

74 less sure of themselves than were the boys, but more positively oriented towards other people. Teacher ratings ranked the girls as significantly better than the boys in the social orientation dimension at a probability level beyond .01. On the other hand, the teachers rated the boys better in cognition with the difference significant at the .10 level. Another factor that may have been related to differences in teacher ratings in Groups A and B was subject matter. Group B dasses were concerned with science and English grammatical structure, whereas Group A was composed of social studies classes. Based on general observation, a case could be made that the girls responded better, cogni- tively, in social studies than in science and English. Thus, in the cognitive dimension, a better rating for the boys would be expected from the science and English teachers. Overall, the results strongly indicated significaw sex differences in many of the variables included in the present study. Major differences seemed to lie in the attitudinal and self concept dimensions, but attitudes towards self and others played a role in classroom performance. It appears that the observed differences apparently related more strongly to attitudinal and motivational factors than to basic cognitive differences.

Subgroup Differences in Cognitive Styles The highly publicized report of Getzels and Jackson (1962) and repli- cations by Torrance (1959) have generated considerable interest in the "divergent thinker." The present project offered an opportunity to add data to the general question of differences in cognitive style. The procedure used to select the students was slightly different from that used by Getzels and Jackson, and Torrance. In each class section, children were ranked in order of their performance on measures of divergent thinking (Uses and Consequences breadth scores) and on IQ tests. Each class was divided into a top third, middle third, and bottom third in performance on these measures. Students in the top third on IQ scores but in the bottom third on divergence scores wereplaced in the high IQlow divergence group. Similarly, those who scored in the top third on divergence but in the bottom third in IQ were placed in the low IQhigh divergent group. Numerous critiques of the original work by Getzels and Jackson have suggested that any such comparison should indude the high IQhigh divergent group, which should represent the most desirable cognitive pattern. This third group was added to indude youngsters in the top third on both IQ anddi- vergent thinking.

75 .7.717771,11r,,,,F771,,4134,1,9,

These stringent requirements, plus the division onthe basis of sex, resulted in smaller but more defensible subgroups.Divisions of the sample by sex were necessary because differential results wereobtained on that variable. While some evidence exists as to differences in testperformance or teacher ratings of students with various cognitivestyles, no evidence has been related to differential classroom performance.Table 22 shows the relationship of cognidve style and classroom performancein the four primary categories. The significant differences wereobtained by the Mann-Whitney U test (Siegel, 1956), since the small Nand wide variations argued against use of the t test. No significant differences, or differences approachingsignificance, were obtained on any of thecognitive style subdivisions among the boys. In other words, whether the boy was classified as highIQhigh divergent, low IQhigh divergent, or high IQlow divergent made no difference in his performance on anyof the major dimensions. The same cognitive style classification among the girls,however, made a substantial difference. Although there were nodifferences among the three groups of girls oncognitive memory expression, the high IQhigh divergent girls were significantly more expressive than the high IQlow divergent girls in the three areas of divergent, convergent, and evaluative thinking. This appearedreasonable on the basis of the a priori expectation that the high IQhigh divergentgirl would have the best traits of both dimensions and be the more effective student. This was not the case among the boys. In the comparison of high IQlow divergent girls with low IQ high divergent girls only one significant difference was obtained. Sur- prisingly enough, this was not in divergent thinking, as one might expect, but in convergent thinking, with the girls classified lowIQhigh divergent significantly more expressive than high IQlow divergent girls. Another difference was noted between the high IQhigh divergent and the low IQhigh divergent groups: the dimension of high IQ seemed to influence the amount of evaluative thinking. On the basis of these and other data collected in this study, the relationship of cognitive style to classroom and test performance obviously has to be qualified according to sex. Table 23 indicates the differences among the different cognitive style groups based on teacher ratings of cognitive ability and sociability. All groups obtained favorable ratings in both areas. A midpoint score on the scale of cognitive abilities was 15 and amidpoint score on the scale of sociability was 2.5. The mean scores of all the subgroups were considerably lower than the midpoint (the lowerthe score, the more favorable the rating).

76 Cognitive Style Related to TABLE 22 Classroom Performance by C lassroo m High IQ- Boys Adjusted Classroom Scores Low IQ- High IQ- Low Divergent High IQ- Girls Low IQ- ExpressivenessCognitive Mean High Divergent Low DivergentHigh IQ- (N =9) 59.33 (N = 10) 52.10 High Divergent (N = 14) 51.21 High Divergent (N = 10) 54.60 (N = 11) 34.18 High Divergent (N = 14) 47.3637.54 MemoryThinkingDivergent Mean SDSD 47.3766.1165.62 39.0935.6049.50 45.2380.3851.00 48.6877.7887.60 24.9729.9219.18 46:2733.71 ConvergentThinking MeanMean SD 58.4861.0255.56 43.9359.6052.90 66.5045.3348.64 92.9062.5581.60 33.4515.1416.00 44.3947.3634.93* ThinkingEvaluative**Significant*Significant at .01 level at .05 level SD 42.27 49.80 91.49 122.68 53.11 52.34 TABLE 23 Ratings on Differences between CognitiveStyle Groups by Teacher Cognition and Sociability Teacher Ratings \ Cognitive Abilities Socl'ability i. Mean SD Cognitive Style Group Sex N Mean SD

Boys 9 10.44 3.08 2.22 1.09 .81 High4HighDivergentGirls 11 9.00 1.73 1.64 .74 Boys 10 8.10 3.17 2.10 .83 High IQLow Divergent Girls 11 10.91 5.02 2.09 .17 Boys 14 11.21 5.03 2.21 Low IQHigh Divergent Girls 14 11.57 3.58 2.07 .83

Differences beyond .05 level between: High IQHigh Divergent Girls vs. LowIQHigh Divergent Girls onCognition High IQHigh Divergent Girls vs. HighIQHigh Divergent Boys onSociability Differences between .10 and .05 level: High IQLow Divergent Boys vs. LowIQHigh Divergent Boys onCognition High IQLow Divergent Boys vs.High IQLow Divergent Girls onCognition

Nevertheless, the comparison ofsubgroups by the t test revealed certain differences. On thebasis of the work by Getzelsand Jackson, it was thought that oneexpected difference would be a morefavorable rating for the high IQlowdivergent than for the lowIQhigh divergent youngster. As canbe seen in Table 23, thesetendencies occurred in the case of the boys.The difference obtained wasbetween the probability levels of .10 and .05 in thedirection of the more favorable attitude for the high IQlowdivergent boy, and this groupof boys also rated more favorably oncognitive abilities than thegirls in this same category.The boys' ratings on cognitiveperformance seemed as high as possible. At an even more significantlevel, differences were foundbetween the high IQhigh divergentgirls and two other subgroups.They received higher ratings on cognitiveabilities than the low IQhighdi- vergent girls and wererated at a higher level ofsociability than the high IQhigh divergent boys.In general, girls in thepresumed ideal category of high IQhighdivergent were rated highlyby the teacher on both cognitiveand sociability levels; thehigh IQhigh divergent boys were rated less favorably.The high IQlow divergentboys received the most effective rating oncognition while the lowIQhigh

78 Comparison of Cognitiw Style Subgroups on Semantic Differential TABLE 24 Girls High Divergent High IQ- (N = 9) Low Divergent High IQ-(NBoys = 10) High Divergent Low IQ-(N = 13) SD High Divergent High IQ- (N= 11) SD Low Divergent High IQ- (N = 9) SD High Divergent Low IQ-(N = 14) SD FatherMotherConcepts Mean74.3367.2272.89 7.336.515.75SD t 68.60t 66.50 Mean 64.00 11.86 6.809.54SD Mean 68.0869.2372.69 10.576.519.38 Mean 66.8170.1871.18 10.109.065.25 * Mean75-7871.7875.33 2.446.635.12 Mean68.7971.1473.86 10.17 5.668.99 WorkStudentLoveSuccessCompetition 67.1171.4465.11 12.4210.33 8326.75 t 56.70 673062.8068.20 12.3110.4410.09 8.26 * 69.7766.2369.0866.54 12.3910.64 9.517.39 64.5569.9168.0070.36 6.929.537.617.03 73.1170.0069.8968.78 6.076.984.575.91 67.0769.3670.3668.64 10.19 7.349.378.03 TotalSchoolImaginationFaith 68.7466.5666.0069.67 14.77 6.829.016.22 65.30623066.1064.82 10.1512.42 9.687.94 69.4667.7771.62 10.78 9.156.627.22 69.1671.0067.7371.91 11.48 8.428.468.17 68.4475.0072.1973.78 3.945.895.866.26 69.6670.6467.2173.79 10.35 9.527.367.74 tp*p < < .05 .10 ononLowHigh concepts concept IQ-HighIQ-High of of Mother, Work Divergent Father, > High Work IQ-Low Divergent op < .05 High DivergentIQ-Low Divergenton concept >of HighMother IQ- IC rr T7,797V,3T.,WIVI7.77,770,

divergent boys received lower ratings in both cognition and sociability. Table 24 compares cognitive style subgroups on the semantic differential scale. Differences and some consistent trends were noted, despite the small samples. With the boys, there was a significant difference between the high IQlow divergent group and the other two subgroups on the concept of Work. Of all concepts, Work received the lowest mean rating by a large margin for the high IQlow divergent group. The same group was also less favorably inclined on the concepts of Mother and Father than were the high IQhigh divergent group. This difference was obtained only at the .10 level of significance but was surprising in view of the expectation that the high divergent children would be more negative toward school and traditional elements in society, including family members. These expectations were not realized with the low IQhigh divergent groups either. No differences were obtained between the high IQhigh divergent and the low IQhigh divergent groups. For the boys, the characteristic of divergence, whether low or high, determined the few differences obtained. Of the ten subscales, the high IQlow divergent group obtained a higher mean score than the low IQhigh divergent group on only two concepts, Competition and Love. Perhaps this fits into a general pattern: high IQlow divergent boys see the world as a rather harsh place where competition thrives and where hard work is necessary to reach one's goal. However, one does not have to like work as a consequence. In comparison, only one difference was found among subgroups of girls. The high IQlow divergent group was significantly higher in ratings of Mother than were the high IQhigh divergent groups. The high IQlow divergent group tended to have slightly higher mean scores than the other two groups on other concepts. No differences were obtained in comparisons of high IQhigh divergent and low IQ high divergent girls. it can be concluded that the minor differences on the semantic differential centered around attitudes toward parents and towards the concept of Work. Other presumed relevant concepts such as School, Imagination, Student, and Success brought forth no important differences.

Differences between Expressive and Nonexpressive Students Since the component analyses suggested that only the one component was related to dassroom expressiveness, an attempt was made to differentiate students on this basis by comparing the top third of the sample with the bottom third in classroom expressiveness on a number of

80 different variables. Expressive and nonexpressivestudents were compared on cognitive factors and teacher ratings(Table 25). In general, differences foud on measures of cognitiveperformance were small or inconsequential.There was a difference at the.10 level of confidence on verbal IQ, withthe more expressive boystending to obtain a higher verbal IQ score than the nonexpressiveboys. With the girls, however, no differences werefound on this variable. Although the tradit'onal levels of statisticalsignificance were not reached on the solutionsscore on theConsequences test, there was a difference in direction. The most expressiveboys gave the greater number of solutions on the Cosequences test,while the most expressive girls gave fewer solutions thatheir nonexpressive counterparts. In the area of teacher ratingsof cognitive ability, significantdifferences were foundbetween the expressive and nonexpressive forboth the boys and girls. Expressive youngsters were givenhigher ratings on cognitive ability. It can be argued,f course, that this very expressive-

TABLE25

Comparison of Expressive and NonexpressiveGifted Students on Cognitive Factors and Teacher Ratingsby Sex Expressive Nonexpressive SD Variables Sex N Mean SD Mean

Verbal IQ Boys 27 133.001- 9.56 7 128.44 7.93 9.36 Girls 27 128.22 10.46 29 127.24 Nonverbal IQ Boys 27 129.33 7.62 27 129.89 8.45 Girls 27 128.81 11.14 29 127.69 11.21 Uses Breadth Boys 26 21.58 6.46 27 19.67 5.53 Girls 27 19.96 8.03 29 17 48 5.67 Consequences Boys 26 15.00 3.90 27 14.11 3.47 Breadth Girls 27 13.78 2.65 29 13.48 2.89 19.22 Consequences Boys 26 39.88 18.81 27 31.78 7.82 Solutions Girls 27 41.04 20.05 29 47.52 Teacher Rating Boys 26 8.65** 3.39 27 11.74 4. 11 (Cognition) Girls 27 10.56* 3.77 29 12.86 4.57 Teacher Rating Boys 26 2.31 .79 27 2.41 .93 2.00 .76 (Sociability) Girls 27 2.09 .62 29

fSignificant at .10 level *Significant at .05 level **Significant at .01 level

81 ness in class influenced the teacher's judgment. However, the teacheT's favorable rating did not carry over into the sociability dimension. No differences were found between the groups on social success and no trends noted. Overall, the differences found on cognitive factors between expressive and nonexpressive youngsters were not very impressive. Expressive and nonexpressive students were compared on the semantic differential scale (Table 26). Again, substantial sex differences were noted. With the girls, differences were obtained on only one item

TABLE 26

Comparison of Expressive and Nonexpressive Gifted Students on Semantic Differential Scale Expressive Nonexpressive Concepts Sex N Mean SD N Mean SD

Mother Boys 25 68.84 6.18 26 74.26** 5.41 Girls 25 72.32 5.93 28 72.17 5.18

Father Boys 25 66.36 9.40 26 70.50 7.93 Girls 25 70.16 8.49 28 70.25 7.30 Student Boys 25 63.88 9.71 26 65.34 731 Girls 24 67.12 7.75 28 67.89 7.26

Work Boys 25 63.60 6.54 26 67.73t 8.20 Girls 25 66.56 7.87 28 64.92 8.66

Competition Boys 25 66.28 9.32 26 67.61 13.77 Girls 25 69.04* 6.57 28 64.53 8.85

Success Boys 25 59.72 11.76 26 70.84** 8.07 Girls 24 66.83 9.20 28 67.50 8.58

Love Boys 25 66.76 9.81 26 70.65 9.02 Girls 25 69.80 7.07 28 68.42 7.18

Faith Boys 25 67.72 9.14 26 72.11 9.06 Girls 25 76.66 8.29 28 73.17 7.60

Imagination Boys 25 66.60 7.31 26 32.15 14.72 Girls 25 68.96 9.63 28 65.85 7.39

School Boys 25 65.72 10.40 26 71.69* 7.20 Girls 25 70.24 7.10 28 68.46 732

tSignificant at .10 level *Significant at .05 level **Significant at .01 level

82 -Competition. This suggestedthat the girls with betterfeelings toward competition are more able tointeract verbally in theclassroom. With School, boys, differences werefound on the conceptsof Mother, Success, and Work, all in the samedirection, with thenonexpressive students producing significantly higherratings. Since the meanneutral score rating these was 49, itcould not be said that theexpressive group was concepts negatively,but they were less positivein attitudes. One plau- sible explanation of these sexdifferences is that expressiveboys have more selfconfidence and self assuranceand thus are not inclined to Possibly students less give extremely positiveratings on these concepts. judgments sure ofthemselves and less inclined tomake independent rate concepts such asMother and School highlypositive. Table 27 shows thecomparison betweenexpressive and nonexpressive ttudents in performance onthe sentence completion test.A low score represents apositive performance. Withthe girls, two differences self concept were found:the expressive girls had(a) a more positive and (b) a more positiveattitude toward conceptssuch as Imagination, Coctpetition, and Creativity.This result supports thenotion that the

TABLE 27

Comparison of Expressive andNonexpressive Gifted Students on the Sentence CompletionTest

Sentence Expressive Nonexpressive Completion N Mean SD Variables Sex N Mean SD 4.15 27 20.92 4.16 Self Boys 24 21.41 Girls 24 20.96t 4.13 29 23.21 4.13 27 3.74 1.20 Fears Boys 24 3.91 1.53 Girls 24 3.88 L15 29 4.28 1.25 5.96 1.56 Time Boys 24 6.50 2.22 27 Girls 24 5.67 1.63 29 5.68 1.91 10.74 2.86 Work Style Boys 24 12.00 3.45 27 Girls 24 10.50** 2.90 29 12.89 3.06 27 11.15 3.74 Family Boys 24 13.41** 3.33 Girls 24 10.92 3.56 29 12.44 3.74 14.30 2.44 Other People Boys 24 14.20 4.29 27 Girls 24 13.25 2.29 29 13.62 2.99

t Significant at .10 level Significant at .05 level Significant at .01 level

83 expressive girls possessed more positive self concepts and self assurance, enabling them to perform effectively in class even when expressiveness was not part of the perceived feminine role. Only one difference was found in the boys, again consistent with the semantic differential results. The expressive boys were not as positive in their attitudes toward their families as were the nonexpressive boys. There wasa hint that dependency relationships in the nonexpressive boys may hinder them from entering into the competitive environment of classroom interaction. Although differences on family ratings were not significant, the direction of differences for the girls was the reverse of that for the boys, with expressive girls leaning in a more positive direction toward their parents. This supported the idea that it is the well adjusted, well rounded girls who participate in classroom interaction.

Family Data As part of the overall project investigating the productive thinking of gifted students in the classroom,an associated study was conducted on the relationship between parental attitudes and the classroom behavior of these gifted students Uenné, 1965). Comprehensive questionnaires were sent to the parents of the students. Each parent was asked to complete the questionnaire independently of spouse. Eighty-seven percent of the mothers and 80 percent of the fathers completed the questionnaires. The questionnaire asked for data on demographic variables suchas number, sex, and birth order of children; age of family members; date of marriage; education of parents; religious preference; father's occupational history; and mother's employment since marriage. Informationwas also obtained on parental involvement with the school (e.g., PTA attendance, extent of personal contact with teachers), attitudes on child rearing, family in- terpersonal relationships, family values, etc. This information was then related to the classroom and test data. Farber's (1962) Index of Marital Integration was used to provide both a quantitative measure of marital integration and an indication of quali- tative aspects of family organization. To assess parental child training practices, particularly achievement inducing and independence grant- ing, a modified version of Torgoff's (1960) Developmental Timetable was used. The Life Goals and Behavior Choices instruments were used to assess parental aims and goals relating to socialization of the child. Few clearcut relationships were found between parental attitudes and classroom behavior or test performance. One trend noted was that parents of divergent boys appeared more controlling, while the parents of divergent girls were less controlling than average for this sample.

84 This finding was tentativelyinterpreted as suggesting that inorder to stimulate maximum divergentthinking, parents must be inconflict with the prevailing norms.This difference would bemarked by their encouraging more constructivelyaggressive behavior in girls andby channeling, through greatercontrol, the aggressive behavioralready permitted to boys. The lack of substantialrelationships between the two setsof variables, however, may have beendue to the relative homogeneityof the samples (predominantly highlyeducated, middle class) or the problems with instrumentation(attitude measurement being aninexact science, to say the least); or it mayreflect a true situation, i.e., that there really are few relationshipsbetween parental attitude and this kind of child behavior. Furtherresearch is needed to clarify theinter- esting questions posed.

85 5 Discussion and Implications

The results of this project haveimplications along a number of dimensions: cognitivestructure and process, cognitive style, the development of measuring instruments,sex differences, and the teaching environment of the classroom. Eachof these will be discussed.

Cognitive Structure and Process The experience of the projectstrongly supported the need fora theoretical model from whichto organize and process information. With- out a model such as Guilford'sstructure of intellect it would have been difficult to makeany sense of the vast amount of complex information gathered. At thesame time, the use or application of sucha model often raises certaindoubts about the completenessor appropriateness of the model itself. Theoriesof behavior developed in the laboratory depend to a largeextent on how well they deal with themore complex relationshipsoutside of the laboratory. Guilford(1958) fully recognized such complexityin discussing the creativeprocess.

No creation takes placein a vacuum. Wecannot create unless we have the background of informationwith which to create. This impliesa history of cognitions and ofmemory for those cognitions. Furthermore,to be come pletely effectiveas a creative person, one must exertsome degree of evaluation. Too much evaluationapplied too early is, ofcourse, detrimentalbut there is one special roleof evaluation that must be pointedout. The act of evaluation leavesus satisfied with our results or it leavesus dissatisfied. If we are too easily satisfied we miss opportunitiesfor making important corrections and improvements and thegeneral opportunity to grow (p.6).

As noted earlier, doubtswere raised about the legitimacy of claim- ing that different mentaloperations were required for thedimensions of convergent and divergentthinking. For example,a student is shown a balloon in a bell jar. The laboratory assistant manipulatessome type of valve and the balloon alternatelygett larger and smaller.

86 Convergent Question Divergent Question Why does the balloon get smaller? How many different solutions can you produce to explainwhat you see?

Student Answer Student Answer If the balloon's size is determined Air is being pumped out of the bell. by pressures internal vnd external A heavier gas is being pumped to its surface inside the balloon. And the balloon is changing size It is an optical illusion. Then the pressure on the outside of There is another balloon within the balloon must be changed first balloon that is being Therefore the lab assistant must be manipulated. manipulating air pressure within the bell jar.

It should be clear that each of the answers tothe divergent questions are merely the "therefore" statementsof a logical syllogism of the same type that was spelled out to answerthe convergent question. The mental operation the student mustperform with the information is of precisely the same order in bothquestions, but on the divergent question he is asked to do it more than once.In a very real sense, the responses to divergentthinking questions are products for whichthe mental process has been inferred. A further implication is that, to increasethe total output in the divergent situation, less likelysolutions will have to be sought. If there is a difference between the two categories,the difference lies in the strategy of seeking manydifferent answers to the divergentquestion, but this is less a mental operation onthe available information than it is a search for alternatives. As notedbefore, this suggests that divergent thinking is more a style than a separatemental operation. It should also be clear that a sizable amountof prior information regarding the problem is needed by thestudent before he can success- fully deal with the question, in eitherthe convergent or divergent framework. Thus, while memory isindependent from divergent and convergent thinking in theGuilford model, it is rarely or neverin- dependent in the solution to a specificproblem. This is the real reason for the high intercorrelations foundbetween the major thought categories in dassroom expressiveness. The role played by evaluative thinkingshould not be overlooked in the two solutions to the balloon question.In the answer to the convergent question, the appropriatenessof each step must be weighed in

87 the reasoning process to see if it fits previous knowledge. Thus, the individual asks himself questions at each stage of thinking. "Is it true that the balloon's size is determined by pressures on its surface? Per- haps there is another way. No! This is the most reasonable answer. Continue with your reasoning." In the divergent question, a student applies evaluative thinking in determining the number and kinds of responses he will report to the questioner. Consider the followingline of thought: "Perhaps there is a little man inside the balloon alternate- ly stretching and folding his arms. Should I report this as one of my solutions or not?" The evaluative decision not to report such far out solutions will reduce the student's score on divergent thinking. No won- der there is reported a relationship between divergent thinking and mental playfulness! Another way of looking at mental playfulness is that it represents a disinterest in applying strict evaluative standards to the output. The humor of the ridiculous is of a similar stripe. One conclusion of this study is that the model of the structure of intellect is incorrect when commonly pictured as a cube with independent categories of cognition, memory, divergent thinking, convergent thinking, and evaluation across the operations dimension. Even in a laboratory situation, an intimate interrelationship appears among these mental operations, obscured only by the special nature of constructed tests and the particular methods of analysis used to develop the model. Some thought has been given to an alternate model to portray the thinking operations involved in problem solving. Potentially fruitful sources for such a model are analogies drawn from computer operation, even though these analogies have limitations in direct applim- tion. The human thinking apparatus goes through a complex matu- rational process, undoubtedly changing the rules regarding information processing from one developmental stage to the next. Also, the human being is able to change his search techniques when standard methods do not apply. Nevertheless, it seems possible to divide thinking operations into four major dimensions roughly analogous to computer functionsstorage, scanning, matching, and logical operations. Storage refers to information stored by the individual in his complex memory system. This information includes facts, associations, and classes of objects or ideas that are retrievable on demand. Whether a fact is stored as an isolated bit of information or with many rich assod- ations dramatically influences the retrievability of that information. Storage itself has been used as a measure of intellectual ability. Vo- cabulary tests are really measures of the semantic storage abilities of the individual. Schools have often placed heavy emphasis on storage in evaluation of a student's success.

,Z7Z74.7.7.:11 A rather different type of mental operationis referred to by the term scanning, meaning skills used to search forrelevant stored material to solve a particular problem. It can also refer,in an outward sense, to the style of search taken by the individualin seeking further needed information within the environment. Thus, a personfaced with a question, "What is a Heffalump?" and havingscanned his internal storage areas and found little response maythen direct attention to a search for information from dictionaries,encyclopedias, zoos, and children's stories. What has been called divergent thinkingin this study seems to represent a set of rules that govern aparticular type of scanning, as in the tests requesting that large numbers ofpossible answers be given to a proposition or problem. A complex relationship seems to exist between storageand scanning, since part of the information placed in the storageunits of the human computer must be rules whichinfluence or determine the type of scanning for various kinds of problems. This, in turn, creates aset which will influence future information processing. Theinfluence of past input upon future input has beenrecognized by such psychological constructs as perceptual defense, cognitivedissonance, etc. The third major area of cognitive operationcould be labeled matching. In this operation, the individual is comparingvarious alternatives as to similarity or appropriateness to agiven criterion. In the earliest developmental stages, matching refers to recognizingsimple likes and differences. Later, a different kind of matchingis used to determine whether or not a particular object or wordbelongs to a class. Still later in a developmental sequence, it is used todecide whether or not a series of events fits complexcriteria of effectiveness. In this respect, matching closely resembles what is called evaluativethinking in the present study. Evaluative thinking canbe performed for its own sake; it also can be involved in the problem solving aspectof the fourth major operation. The fourth major operation refers to a general systemof thinking under the heading of logical operations. Theindividual takes information given to him and applies certain systematicrules to generate a solution or conclusion. The development of systemsof rules is described effectively in the work of Piaget (Flavell,1963). Thus, while the type of problem depends upon the scanningoperation, the procedure by which the individual generates aconclusion through logical thinking is termed a logical operation. Asused in this study, convergent thinking represents mostclosely the area of logical operations,although the category of divergent implicationsand synthesis is also appropriate. At the present time, it is the opinionof the investigators that all

89 ,g-mx,mrrnwmirrrwurripp,_

thinking operations can be included in these fourcategories to provide the basis for further analysis of data. Cognitive Style. One of the possible links to be forgedbetween the domains of personality and cognition lies in the recentrebirth of interest in cognitive style. Cognitivestyle may be defined as a characteristic set or way in which an individual responds to orapproaches his environment. This includes characteristic ways ofprocessing information. Gallagher (1964a) pointed out that at least fourinvestigators have been looking at cognitive style in terms of thewillingness of the individual to take intellectual risks. Maslow (1956),Schachtel(1959), Witkin et al. (1962), and Getzels and Jackson (1962) haveinvestigated and differentiated these styles. Maslow calls them defenseand growth; Schachtel refers to autocentric and allocentric;Witkin et al. mention field dependent and field independent; and Getzels andJackson differentiate high IQ and high creative. The basis for such styles seems tolie in the personality domain, yet the style reveals itself in cognitive performance. In the present study, the findings oil cognitive style wereinteresting enough to warrant further investigation. If classroom expressivenessis considered a style, some degree of consistency was found in students' expressiveness from one point in time to another, and from one classroom to another. This expressivenessalso seemed to be related to certain attitudes and personality variables. In the replication of Getzels and Jackson's work (1962), a number of expected results were obtained, particularly with the low IQhigh divergent boys. With this group, teachers were less favorable in their ratings on cognition, as predicted, while the high IQlow divergent boys seemed better accepted by the teacher yet showed differencesin attitudes toward concepts, especially Work, to which they responded negatively. These results suggest that while Getzels and Jackson, and Torrance (1959) have been severely taken to task for technical short- comings in their research, they identified some cognitive style variables worthy of further exploration. The general description of the low IQhigh divergent youngster seems to hold, at least for the boys in the present study. There are differences in their attitudes, and teachers regard them less favorably than the youngsters called high IQlow divergent. The implications of these findings for education are something else again. One popular interpretation is that the creative (divergent) student is smothered by the traditional educational setting and discrim- inated against by the conformist teacher. Another interpretation, just as likely, is that the high divergent but not highIQ student is, in fact,

90 unsuccessful as a learner and not particularly adept atmobilizing his scattered intellectual resources to attack a problem.The teacher then is merely calling them as he sees them when he saysthat the low IQ high divergent youngster is not as good a thinker,since he is not, for most traditional classroom tasks. Sex played a role in the differences obtained in cognitivestyle. The high IQhigh divergent girls were found to be significantlysuperior in classroom expressiveness, teacher ratings, and personalitycharacter- istics. The same was not true of the high IQhigh divergentboys. With the girls, high expressiveness and high scores on variousinstruments accompanied a general pattern of good academic adjustmentand acceptance of the academic role. In theboys, there seemed to be more confusion and overlappingof roles. Some boys were expressive, apparently for academic gain, while others were expressive forself en- hancement. Obviously, similar behavior may imply quite different motivational structures in boys than in girls. More thoroughinvestigation is needed. Sex Differences. An impressive literature has beendeveloped over the last decade on the importance of sex as a variable influencing many aspects of behavior. The general results of the presentstudy seem to be in harmony with previous work. The gifted boys seemed to enjoy more self confidence in their own abilities, while the gifted girlsseemed to express more positive attitudes towardother people and concepts such as Mother, School, and Work. Thesedifferences in attitudes were probably established in early childhood. As Kagan (1964) pointed outin a recent review:

The child as young as four has dichotomized the world into male andfemale people and is concerned with boy-girl differences. By the time heis seven, he is intensely committed to molding his behavior in concordancewith cultural standards appropriate to his biological sex and he shows uneasiness,anxiety and even anger when he is in danger of behaving in waysregarded as characteristic of the opposite sex (p. 162).

Evidence has accumulated showing a variety of applicationsof this general theme. Torrance (1963) has shown that thenumber of ideas produced to improve children's toys was dependent onthe sex identification of the toy in question. Primary school boys gave moreideas on improving a toy if it was seen as masculine and vice versa forthe girls. Crandall and Rabson (1960) found that boys were more likely than girls to return to unsolved puzzles, and to be more analytic, independent, and persistent in their approach to problems. Whether or not such differences are genetically or environmentallybased, or repre-

91 sent a subtle blend of both influences, is less important for educational action than recognition that differences exist. Although the girls in the study seemed on a par with the boys in written performance, they were less expressive in the classroom. It seems reasonable to conclude that the sex role identification of die girls had an inhibiting effect on their classroom expressiveness. As Kagan (1964) pointed out:

It is not uncommon for an adolescent to view intense intellectual strivings as a competitively aggressive behavior. In order to obtain the best grade in a class, one must often defeat a peer in open debate and in examination scores. The competition has obvious aggressive overtones. Since the typical female has greater anxiety over aggressive and competitive behavior than the male, she experiences greater conflict over intellectual competition (p. 158). If oral expressiveness is viewed as a positive trait for girls, it is clear that special steps will need to be taken to encourage it. The Measurement of Creativity. Some of the measuring instruments used in the present study have been labeled creativity tests by other investigators. This is an overenthusiastic term, to say the least. It is like describing one piston movement in the gasoline engine and saying that one has captured the essence of the entire cycle. Yet there seems little doubt that these tests are measuring something that is different from what is measured by standard IQ tests. It was suggested earlier that this something was a type of cognitive style. As other investigators found, a small correlation was noted between IQ and divergence, or fluency, or breadth, or whatever term may be given to the production of a large number of answers to a given question or proposition. Furthermore, this characteristic seemed reasonably stable within individuals. The best estimate of the project staff was that scores on divergent tests, particularly scores which take into account only total production, are really measuring one component, intellectual risk taking. The role of evaluative thinking in such tests needs to be thoroughly investigated. The production of a long string of answers to questions such as, "What would happen if everyone lived to be 200 years old?" requires a certain suspension of judgment. Each answer projectedcan stir up a raft of self criticism with attendant objections or qualifications. Unless the person is willing to put these aside and go on to the next idea, he will not be productive on a timed task. If it is true that temporary suspension of critical judgment is one necessary component for creative thinking, we can see how tests of divergence have a firm grip on one part of the mystical elephant, creativity. The new idea or original proposal tends to wobble and flutter intellectually like a wounded quail and hasn't a chance against academic

92 4,7,777,9',7:7017,77.777.W.7,773515774.,

marksmen whose great joy lies in bringing fledgling ideas down to earth with a sickening thud. Of course, the joke is on the critics, since even- tually they shoot down their own new ideas with the same ferocity and are unlikely to produce new contributions of any consequence. On the other hand, it is unlikely that the compulsive free association expert, who has dozens of ideas for every problem but cannotfollow any one of these to conclusion or bringhimself to evaluate which of many proposals might work, will producemuch more than the critic who is always evaluating. The divergent tests may very well predict high creativity for such a person and be wrong, for the creative person must apply evaluation as well as suspend it, and the failure todo either can be debilitating to actualproduction. One of the best but least likely things that could happen would be for investigators to avoid assiduously the term creativity in referring to these embryonic measures. The Classroom Environment. It has often been said, and said proudly, that teaching is an art. Indeed it is. It might serve a useful purpose for the education profession to considerin more detail precisely what that statement means. It seems to mean that we can observe, as a result of the efforts of the bestpractitioners, an extremely desirable product, but the means or process by which this product was obtained cannot be determined. The implication is often made that not only has this teaching process not been measured, but that it cannot bemeasured. This viewpoint holds that some kind of mysterious magic creates a fortuitous blend of indescribable ingredients possessed by only a few fortunate practitioners. Training others usually consists of students sitting at the feet of the masters, hoping in some unpredictable way to absorb the magic skill. This attitude, of course, is observed not only in the education profession. At various times, it has been stated that medicine is an art, psychotherapy is an art, etc. The individual brilliance of some practitioners of such arts in the past has not blinded these professions tothe necessity of changing from an art to a science as soon as possible. De- spite the brilliance of certain individual physicians, good medical service for most people has relied on scientific advances in anumber of disciplines. The situation in education seems parallel. Blessed with a teacher who has an intuitive knowledge of educational process, a few students receive manifest benefits. However, what about the other students, most of whom cannot be expecled to meet thisone-in-a-thousand gifted teacher? Our responsibility would seem to lie in making education less of an art and more of a science, by studying the processesby which desirable cognitive products are obtained.

93 One of the most powerful observations made as a result of thedata collected in the present study was how central and crucial theteacher's role is as initiator and determiner of the kinds of thought processes expressed in the classroom. It was the teacher who asked thedirecting questions which were the focus of the discussion. The way inwhich these questions were presented determined the kind ofthought operations the students performed. The tendency of the student torespond in the same idiom as the teacher was so strong thatwhenever it did not occur, the observer was leftwith the feeling that either the student did not understand the question, orsomething was grievously amiss in the rules governing teacher student operation in that classroom.A student, for example, could not respond to a questionsuch as "Who were the three best presidents the United States has everhad?" without saying something that approached evaluative judgment.It is inconceiv- able to consider a student responding effectively to"What is the formula for nitric acid?" without giving a factual memory response.The same shaping role playedby teacher questions has been noted by Taba et al. (1964).

The questions teachers ask set the limits within which students can operate and the expectations regarding the level of cognitive operations. Questions are the carriers of whatever new cognitive system is emerging. Somequestions function as invitations to heighten the performance of certain cognitive operations, while leaving the content in the direction of these operations open.Such questions invite invention, discovery, the creative use of previousknowledge. Others control and limit both the content and nature of cognitive operation (P. 177). These observations place both a great responsibility and great opportunity in the hands of the teacher. It means that the teacher can direct, by questions, the flow and level of thought complexity in the classroom. It is our conclusion that this skill can be acquiredby teachers if training is given in the system, and practice allowed inits application. There is little doubt in the minds of investigators who have studied the teacher and the classroom that it is possible to measure,modify, and improve the teaching process. Medley (1963), discussing the OS- cAR technique (observation schedule and record), made the following statement:

Perhaps the most important conclusion to be drawn from these two studies, is that meaningful measures of classroom behavior can be developed from objective records made by relatively untrained observers with a rather crude instrument...(p. 272).

94 Wal len and Travers (1963) suggested that it is possible to construct a model within which one can construct alaw of teacher behavior. Their formula is: T f (R1,111). T represents the behavior of the teacher, which is a function of the goals to be achieved,;, and the present behavior of the pupils, Rt. On thebasis of experience in the resent study, the means by which the goals areobtained, as well as the goals themselves, are important topics.

Project Limitations Every research project has built-in iimitations which restrict theinter- pretation of results and circumscribe the number and kind of inferences that can be drawn. In this study, many of the limitations arefairly obvious. Sample of Subjects. One of the factors limiting the results wasthe type of subjects upon whom the data wascollected. These students were intellectually gifted, with a history of successfulacademic achievement. There were no children of average or below average mental ability, nor were there intellectually superior underachievers.This limitation probably influenced the finding that IQ score was not a significant variable and may have been responsible for some of thelimited relationships obtained between personality and attitudinal variables and classroom performance. One would expect that themajority of these youngsters would be moderately well adjusted, and thus there would be no negative extreme along the continuum of emotionaldistur- bance and maladjustment. Age of Subjects. The sex differences found may very well be specific to the age level of the youngsters studied. Between the agesof 12 and 16 there emerges a greater consciousness of sex role and this changecould well have influenced and biased the results. It should notbe assumed that similar results would be obtained on gifted, intermediategrade children or on senior high school students. Teacher Sample. Any generalizations from this studyregarding teacher behavior are limited by the small number of teachersinvolved and the fact that these teachers were specifically chosen towork with gifted students. An unselected group of teachersprobably would show greater range and variance. Since the sample groupshowed impressive variance on most measurable characteristics, theconclusion is that teachers' performance, as categorized in thestudy, shows much variation from one teacher to another, although there is someconsis- tency within the individual teacherperformance. Limitations of Classroom Recordings. The recorded sampleof teacher and student behavior must be considered only one component

, of the whole learning environment. Not all, nor necessarily the most important, interactions between teacher and student occur in the classroom discussions. Perhaps the best teaching is done inindividual con- ferences with students, through criticisms and suggestions in development of individual reports, or in the stimulation of personal research projects. While it is tempting to assume that the teacher style in these group discussions would imply certain kinds of behavior inindividual contacts, this is not necessarily so and would have to be demonstrated. Measuring Instruments. Every research study is, in the end, at the mercy of the measuring instruments. This project is no exception.One of the key instruments was the classificadon system devised as part of the total project. The classification system itself appears to be a potentially useful tool in describing teacher and student behavior and in categorizing differences among students and among teachers. One of the problems noted early by the judges was that the system might be too atomistic, too refined in its ratings to show accurately what was going on in the classroom. Thus, the detailed rating of each statement made by teachers and students sometimes detracted from the larger meaning and may, as a consequence, have been somewhat misleading. In the future, larger units of measurement should be developed to describe more adequately classroom interchange. In addition to the classification system, the results were dependent upon tests of attitude, such as the semantic differential and sentence completion, :Lnd new tests of cognition, such as the divergent thinking tests, which attempt to measure a rather new dimension in intellectual performance. The reliability and stability of each of these measures has been questioned. The risk taken by the investigators was that, by using some measuring instruments which appeared to touch upon important variables, even though they had some technical weaknesses, real differences might have been obscured by the unreliability of such instruments.

Future Research Plans In a program of research, the first study often serves its most important function in making clear what should have been studied if only the investigator had known at the beginning what he knew halfway through the first research problem. Although much useful information was gained through this project, the feeling persists that further analysis along a broader conception of classroom interaction, with improved measuring instruments, can bring forth greater returns in tams of insight into teacher strategies and the teaching process. Further investigations have stressed larger units called topics, which can be identified and classified in a broader system. Hopefully, addi-

96 tional analysis of these classroom topicswill provide even more meaningful information for teacher education. Not all teacher or student behaviors areequally important to the conduct of the classroom or to theadvancement of curriculum goals. There appear to be critical incidents,certain choice points, certain crossroads, in the sequence of theclassroom activities. The decisions at these points have ashaping effect on the classroom operation. An attempt to locate suchchoice points is one logical extension of the present study. Informal contentanalyses of the tapescripts have already been undertaken. It is hoped that aclassification of these various choice points will fit them into the larger topics. Another categorization, the range ofpossible teacher choices in a given situation, and an attempt to explorethe consequences of each choice may pay richer dividends for teachereducation. At first, this would be done clinically by taking specificexamples from tapescripts. Although the results within the classrooms wereonly suggestive, there were hints that certain kinds of cognitive processes arecalled for more often in certainsubject areas than in others. In science,the emphasis seemed to be on the close analysis ofspecific problems. Little attempt was made to explore alarge number of different intellectual paths; instead, emphasis was placed onthe analytic efficiency by which a person follows one line of thought.This seems to call for a review of teacher strategy, if, indeed, othercognitive goals are considered important. The scientific technician mustbe able to follow standard procedures, but the creative scientist mustexplore. A close analysis of a random sample or typical sampleof effective teachers in a given content area would provide ageneral picture of expectations incognitive expression. The sex differences in the presentstudy underline the importance of understanding the attitudes, values, andperceived sex roles of students in order to provide an effectiveeducational environment for them. The results were sufficiently dramatic to suggestthat a major research effort should be made in study of giftedgirls. One line of investigation could compare public versus privateperformance on intellectual tasks. It seems obvious that intellectualaggressiveness is generally accepted as a male role. Girls who areable to compete in this area appear, onthe basis of our present findings, to be the most mature, secure,and well adjusted girls. Does the inhibition ofintellectual aggressiveness in public performance affect the privateperformance as reflected in term papers, individual researchprojects or compositions, or in otherdimen- sions which allow a certain amountof social anonymity for the student? A further study of attitudes andvalues of gifted girls would lead to a better understandingof their problems in modern society.Since gifted girls represent the largest single source of untapped intellectualpower in our society, it seems important to devotemore attention to analysis of their unique adjustment problems at all levels of the educational system. MacKinnon (1962) and others have pointed out the seeminglypar- adoxical finding that creative members of bothsexes appear more like the opposite sex in personality characteristics and attitudes. That is, creative men appear more feminine, while creativewomen appear more masculine. It seems likely that such findings relate to the general characteristics of openness to experiences, as opposed toa closed system where certain experiences and thoughts are denied the individualon the grounds that they are inconsistent with the expected self imageor concept. This leads into the topic of cognitive style,or styles of information processing, and the impressive degree of individual difference in such styles. What family and social variablesare relevant for creation of various styles? How do different types of instruction affect the growth or delay of these styles? The most stringent limitations are set byper- ceived sex role. One notes the contempt society holds formen who engage in the fine arts or the suspicion held for women who invade the physical sciences. Predominantly other-directed individualsare likely to shut themselves off from experiences identified as belonging to the opposite sex and thus reduce the number of effectiveconcepts they can use in problem solving or creative thinking. Knowing the patterns of child rearing and later social experiences which result in the other- directed versus inner-directed adults could provide the basis for allowing greater creativity to the individual and the culture. Despite the emphasis in the current studyon interaction and complex social variables, the role of the laboratory in studyingsome aspects of the problems should be recognized. A prime example is the thinking operation or the thinkingprocessitself.Itis unlikely that much more will be derived from extended analysis of studentre- sponses to questions such as, "How many ways are there to use a brick?" or, "What would happen if pills were substituted for food?" Between the question and answer, much thatgoes unstudied takes place within the mind of the respondent. Intensive analysis of the chain of thinking between question and answer would bemore to the point. New techniques of computer simulation of thinking operationsmay be necessary to find out more about suchprocesses as "insight." If insight represents a generalization about certain data, perhaps thereis a critical mass(say 70 percent of necessary data dealing with the problem) which must be available to conscious manipulationbefore the insight comes. Such topics aremore open to computer study than

98 with human subjects since amountand kind of input to the computer can be controlled.

A Final Word Many social scientists trained inexperimental and research techniques yearn to have a neat,precise problem where a specific treatmentis ap- plied, appropriate control groups areavailable, all extraneous factors either eliminated or controlled, etc.Even the thought of studyingthe multitude of interacting factors inthe classroom must bring afeeling of despair. If the results are dependent onthe teacher, the phase of curriculum development, the personalityand intellectual makeup of the group, numerouschild developmental variables,and the time of the year, how meaningfulis the talk about a controlledexperiment? Perhaps the closest approach tothe problem would be to use one group as its owncontrol and systematicallyintroduce teacher variation such as different types of questionasking behavior. While this would use the class andteacher as its own control, evensuch a simple plan is at the mercy of a numberof variables, to say nothingabout the gen- eralizability of the findings. Perhaps, at this point in time,the model for the social scientistwho wants to study groupinteractions in complex socialsituations should be closer to that of the astronomerthan the physicist. The astronomeris doomed to observe the universeand draw inferences fromobservations without being able to control ormanipulate the variables. The physicist has opportunities to controland manipulate and is notburdened with such matters as therepresentativeness of the sample of matterin his experiment. The rationale for studyingthe classroom does not rest inpossibili- ties for elegant research design;rather, these are important mattersin- fluencing present and futuregenerations. Each scientific area must work out its own destiny, its ownmethods, and its own theoretical models without blindlyimitating older and more maturedisciplines. References

Anderson, H. H. The measurement of domination and of socially integrative behavior in teachers' contacts with children. Child Develprn., 1939, 10, 73-89. Aschner, Mary Jane. The language of teaching. Teachers Coll. Rec., 1960, 61, 242-252. Aschner, Mary Jane; Gallagher, J. J.; Perry, Joyce; Afsar, Sybil; Jenne, & Faar, H. A system for classifying thought processes in the context of classroom verbal interaction. Urbana, Ill.: Univer. of Illinois, 1965. Bloom, B. S.(Ed.) Taxonomy of educational objectives. Handbook I: cognitive domain. New York: Longmans, Green, 1956. Christensen, P. R., Guilford, J. P., & Wilson, R. C. Relation of creative responses to working time and instructions. J. exp. Psychol., 1957, 53, 82-88. Corman, B. R. Learning II: problem solving and related topics. Rev. educ. Res., 1958, 28, 459-467. Crandall, V. J.. & Rabson, A. Children's repetition choices in an intellectual achievement situation following success and failure. J. genet. Psychol., 1960, 97, 161-168. Duncan, C. P. Recent research and human problem solving. Psychol. Bull., 1959, 56, 397-429. Farber, B. Marital integration as a factor in parent child relations. Child Develpm., 1962, 33, 1-14. Flanders, N. A. Teacher influence, pupil attitudes, and achievement. Minne- apolis: Univer. of Minnesota Press, 1960. Flanders, N. A. Intent, action and feedback: a preparation for teaching. J. teacher Educ., 1963, 14, 251-260. Flavell, J. H. The developmental psychology of Jean Piaget. Princeton, N.J.: Van Nostrand, 1963. Gage, N. L. A method for "improving" teacher behavior. J. teacher Educ., 1963, 14, 261-266. (a) Gage, N. L. Paradigms for research in teaching. In N. L. Gage (Ed.), Hand- book of research on teaching. Chicago: Rand McNally, 1963. Pp. 94-141. (b) Gallagher, J. J. The tutoring of brain injured mentally retarded children. Springfield, Ill.: Charles C Thomas, 1960. Gallagher, J. J. Productive thinking. In M. L. Hoffman & Lois Hoffman (Eds.), Review of child development research. Vol. 1. New York: Russell Sage Found., 1964. Pp. 349-382. (a)

100 40,V

Gallagher, J. J. Teaching the gifted child. Boston:Allyn g: Bacon, 1964. (b) Getzels, J. W., & Jackson, P. W. Creativity andintelligence. New York: John Wiley, 1962. Gold, M. J. Divergent production performance as affectedby type of task and length of test. Unpublished doctoraldissertation, George Peabody Coll. for Teachers, 1963. Guilford, J. P. Creativity. Amer. Psychologist, 1950, 5,444-454. (a) Guilford, J. P. Fundamental statistics in psychology andeducation. (2nd ed.) New York: McGraw-Hill, 1950. (b) Guilford, J. P. The structure of intellect. Psychol. Bull.,1956, 53, 267-293. Guilford, J. P. Creative intelligence in education. LosAngeles: Los Angeles County Office of Superintendent of Schools, Div.of Res. & Guidance, 1958. Guilford, J. P. Three faces of intellect. Amer. Psychologist,1959, 14, 469479. Guilford, J. P. Intelligence: 1965 model. Amer. Psychologist, 1966,21, 20-26. Guilford, J. P., Merrifield, P. R., & Cox, Anna. Creativethinking in children at the junior high school levels. Report No. 26. LosAngeles: Univer. of Southern California, Psychol. Lab., 1961. Heyns, R. W., & Lippitt, R. Systematic observational techniques.In G. Lindzey (Ed.), Handbook of social psychology. Cambridge, Mass.:Addison-Wesley, 1954. Pp. 370-404. Hiler, E. W. Analyses of patient-therapist compatibility.J. consult. Psychol., 1958, 22, 341-347. Hiler, E. W. Sentence completion test as a prediction ofcontinuation in psychotherapy. J. consult. Psychol., 1959, 23, 544-549. Hughes, Marie. Development of the means of the assessment ofthe quality of teaching in elementary schools. Salt Lake City: Univer. of UtahPress, 1959. Hutchinson, W. L. Creative and productive thinking in theclassroom. Unpub- lished doctoral dissertation, Univer. of Utah, 1963. jenné, W. The families of gifted children. In J. J.Gallagher, Productive thinking of gifted children. Cooperative Research ProjectReport No. 965. Washington: US Office of Education, 1965. Pp.172-289. Kagan, J. Acquisition and significance of sextyping and sex role identity. In M. L. Hoffman & Lois Hoffman (Eds.),Review of child development research. New York: Russell Sage Found.,1964. Pp. 137-168. Kaiser, H. F. The varimax criterion foranalytic rotation in factor analysis. Psychometrika, 1958, 23, 187-200. Kaiser, H. F. Formulas for component scores.Psychometrika, 1962, 27, 83-87. Kaya, Esin. Developing a test of creativethinking in young children. New York: New York Univer., 1960. (Mimeographed) Kettner, N. W., Guilford, J. P., & Christensen,P. R. A factor analytic study across the domains ofreasoning, creativity and evaluation. Psychol. Monogr., 1959, 73. Maccoby, Eleanor. Developmentalpsychology. Annu. Rev. Psychol., 1964, 15, 203-250. McGuire, C., Hindman, E., King, F., &Jennings, E. Dimensions of talented behavior. Educ. psychol. Measmt, 1961, 21, 3-38.

101 tarlt.

MacKinnon, D. W. What makes a person creative? Sat. Rev., 1962, 69 (1), 15-17. Maslow, A. H. Defense and growth. Merrill Palmer Quart., 1956, 3, 37-38. Maslow, A. H. Creativity in self-actualizing people. In H. H. Anderson (Ed.), Creativity and its cultivation. New York: Harper, 1959. Pp. 83-95. Medley, D. M. Experiences with the OSCAR technique. J. teacher Educ., 1963, 14, 267-273. Medley, D. M., & Mitzel, H. E. Measuring dassroom behavior by systematic observation. In N. L. Gage (Ed.), Handbook of research on teaching. Chicago: Rand McNally, 1963. Pp. 247-328. Osgood, C. E. Studies on the generality of effective meaning systems. Amer. Psychologist, 1962, 17, 10-28. Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. The measurement of meaning. Urbana, Ill.: Univer. of Illinois Press, 1957. Payne, A. F. Sentence completions. New York: N.Y. Guidance Clinic, 1928. Perkins, H. V. Classroom behavior and underachievement. Amer. educ. res. J., 1965, 2, 1-12. Pierce, J. W., & Bowman, P. Motivation patterns of superior high school students. In The gifted student. Cooperative Research Monograph No. 2. Washington: US Office of Education, 1960, 35-66. Rhode, A. F. The sentence completion method. New York: Ronald, 1957. Rotter, J. Word association and sentence completion methods. In H. H. An- derson & Gladys Anderson (Eds.), An introduction to projective techniques. New York: Prentice-Hall, 1951. Pp. 279-311. Ryans, D. G. Characteristics of teachers. Washington: Amer. Council on Educ., 1960. Ryans, D. G. Teacher behavior theory and researdi: implications for teacher education. J. teacher Educ., 1963, 14, 274-293. Schachtel, E. G. Metamorphosis. New York: Basic Books, 1959. Siegel, S. Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill, 1956. Smith, B. 0. A concept of teaching. Teachers Coll. Rec., 1960, 61, 229-241. Smith, B. 0. A concept of teaching. In B. 0. Smith & R. H. Ennis (Eds.), Lan- guage and concepts in education. Chicago: Rand McNally, 1961. Pp. 86-101. Smith, B. 0., & Meux, M. 0. A study of the logic of teaching. Urbana, Ill.: Univer. of Illinois, Bur. of Educ. Res., 1962. Smith, R. Perceptions of self, others, and certain environmental aspects of high and low divergent intellectually superior children. Unpublished doctoral dissertation, Univer. of Illinois, 1962. Spaulding, R. L. Achievement, creativity, and self-concept correlates of teacher- pupil transactions in elementary school classrooms. Urbana, Ill.: Univer. of Illinois, 1963. Taba, Hilda, Levine, S., & Elzey, F. Thinking in elementary school children. Cooperative Research Project Report No. 1574. Washington: US Office of Education, 1964. Tendler, A. D. A preliminary report on a test for emotional insight. J. appl. Psychol., 1930, 14, 123-136.

102 Thorndike, R. L. The measurement of creativity. Teachers Coll. Rec., 1963 64, 422-424. Torgoff, I. Synergistic parental role components: application for expectancies and behavior: consequences for child's curiosity. Paper presented at Amer. Psychol. Ass., Chicago, 1960. Torrance, E. P. Explorations in creative thinking in the early school years.VI. Highly intelligent and highly creative children in a laboratory school. Min- neapolis: Univer. of Minnesota, Bur. of Educ. Res., 1959. Torrance, E. P. Guiding creative talent. Englewood Cliffs, N.J.:Prentice-Hall, 1962. Torrance, E. P. Education and the creative potential. Minneapolis: Univer. of Minnesota Press, 1963. Wallen, N. E., & Travers, R. M. W. Analysis and investigation of teaching methods. In N. L. Gage (Ed.), Handbook of research on teaching. Chicago: Rand McNally, 1963. Pp. 448-505. Withall, J. Research tools: observing and recording behavior. Rev. educ. Res., 1960, 30, 496-512. Witkin, H. A., Dyk, R. B., Paterson, H. F., Goodenough, D. R., & Karp, S. A. Psychological differentiation. New York: John Wiley, 1962.

103